def plot(df: pd.DataFrame, id_vars: list, value_vars: list):
"""
Plots a data frame.
:param df: dataframe to plot
:type df: pd.DataFrame
:param id_vars: list of column names for the x-axis
:type id_vars: list or str
:param value_vars: list of column names for the y-axis
:type value_vars: list
:return: melted data frame
:rtype: pd.DataFrame
"""
if not isinstance(id_vars, list):
id_vars = [id_vars]
data = pd.melt(df, id_vars=id_vars, value_vars=value_vars)
sns.lineplot(x=id_vars[0], y="value",
hue="variable",
data=data)
return data
def generate_plot(data: pd.DataFrame, output_path: str) -> None:
"""
Generates am image plot for the given dataframe
:param data: The dataframe to generate the plot for
:param output_path: THe file output path
"""
palette = {'L': '#875438', 'C': '#0E0C0C', 'U': '#8A8D91', 'R': '#C1A15B', 'M': '#EC7802'}
plt = sns.lineplot(data=data, palette=palette, linewidth=1.5, hue='A')
plt.set(ylabel='Average Proportion of Deck')
fig = plt.figure
fig.savefig(output_path)
df['counts'] = df.groupby('datetime', as_index=False)['datetime'].transform(lambda s: s.count())
x = []; y=[]
for point in df['sentiment']:
x.append(point[0])
y.append(point[1])
plt.figure(figsize=(10, 10))
plt.scatter(x, y, alpha=0.5)
plt.title("Sentiment:Polarity vs. Subjectivity")
plt.xlabel("Polarity")
plt.ylabel("Subjectivity")
plt.show()
plt.figure(figsize=(10, 10))
sns.lineplot(x="datetime", y="counts", data=df)
plt.xticks(rotation=15)
plt.title('Count of Tweet Times By Minute')
plt.xlabel('Time(Minutes)')
plt.ylabel('Count')
plt.show()
df['datetime'] = pd.to_datetime(df['datetime'], format='%m/%d/%y %H:%M').dt.hour
plt.figure(figsize=(10, 10))
sns.lineplot(x="datetime", y="counts", data=df)
plt.xticks(rotation=15)
plt.title('Count of Tweet Times By Hour')
plt.xlabel('Time(Hours)')
plt.ylabel('Count')
plt.show()
def nucleotide_difference_imbalance_plot_stylized_like_Figure_8_of_Morrill_et_al_2016(
sequence_file, chr_, position_corresponding_to_first_nt,
chunk_size = chunk_size, overlap_specified = overlap_specified,
save_vg=False, return_fig=False):
'''
Main function of script.
Make a plot figure like Figure 8, panel B of Morrill et al 2016
(PMID: 27026700), see
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4882425/figure/F8/ .
Inputs:
- sequence file of spanning desired coordinates in FASTA format with
single sequence in it. (Or at least first sequence be the one to use.)
- text string of designation to use for chromosome/contig/scaffold/region in
plot labels
- position corresponding to first bases in provided sequence as an integer.
To be used to label positions along coordinates in x-axis of plot.
Optionally you can provide integer settings for the `chunk_size` and
`overlap_specified` for the analyses windows. Without specifying them, by
default they are set to mirror Figure 8, panel B of Morrill et al 2016
(PMID: 27026700).
You can set `return_fig` to return a plot figure object and would be useful
if calling from a Jupyter notebook or IPython. If you assign it to `ax` in
your notebook, you can redisplay it in another cell via `ax.figure`.
Saving as vector graphics is also an option. It is not the default because
file PNG-style image files are more familiar to most folks and more
convenient in Jupyter notebooks.
Saves an image file of the plot, or optionally, returns a plot object.
'''
# Retrieve sequence from sequence file and break it up into the set chunks
#---------------------------------------------------------------------------
seq_entries = Fasta(sequence_file)
seq = seq_entries[0] # assume first one is the one to be used
chunks = (
list(gen_chunk_string_with_different_step(seq,chunk_size,step_size)))
#discard any chunks at end less than the size of the set window
chunks = [x for x in chunks if len(x)== chunk_size]
# Assign midpoint postions to each chunk based on chunk_size, start position
# input, and provided sequence length.
# ALSO, while going through chunks, might as well do calculation too:
#Calculate G vs. C and A vs. T that I think might give results like Figure 8
# of Morrill et al 2016 (PMID: 27026700)
#---------------------------------------------------------------------------
#determining midpoints for each chunk (relative length of provided sequence)
# and assiging to a list. (used a dictionary in development but a list
# should be better for python 2.7 compatibility)
# Had calculation as separate step in development but why interate again.
# (Doing similar thing where now storing `chunks_diffs` as list and
# not diction for python 2.7 compatibility.)
import sys
#chunks_midpoints = {} #key will be index of chunk in chunks
chunks_midpoints = []
#chunks_diffs = {} #key will be index of chunk in chunks. value will be tuple
# with each diff as an item
chunks_diffs = []
for indx,chunk in enumerate(chunks):
# handle first chunk without much fanfare because easy calculation and
# doesn't depend on a previous one
if indx == 0 and len(chunk)== chunk_size:
#chunks_midpoints[indx] = chunk_size/2
chunks_midpoints.append(chunk_size/2)
elif len(chunk)== chunk_size:
start_curr_chunk = step_size * indx
end_curr_chunk = step_size * indx + chunk_size
#chunks_midpoints[indx]= midpoint((start_curr_chunk,end_curr_chunk))
chunks_midpoints.append(midpoint((start_curr_chunk,end_curr_chunk)))
else:
sys.stderr.write("\n\nError? Issue with size of sequence chunks "
"not matching expected/\n") #shouldn't happen
sys.exit(1)
GC_diff = calc_nt_diff("GC",str(chunk)) # casting 'Sequence' object to
# string with `str(chunk)`
AT_diff = calc_nt_diff("AT",str(chunk)) # casting 'Sequence' object to
# string with `str(chunk)`
#chunks_diffs[indx] = (GC_diff,AT_diff)
chunks_diffs.append((GC_diff,AT_diff))
# Adjust potions of midpoints to account for first base in sequence file
# being not first base of chromosome
#---------------------------------------------------------------------------
# correct chunks_midpoints to take into account that first postion in
# provided sequence might not be first position of that sequence along
# chromosome so that labels for x-axis will match situation
# (This was done with a dictionary comprehension in development where used
# dictionary)
start_pos = position_corresponding_to_first_nt
#chunks_midpoints = {k:v+start_pos for k,v in chunks_midpoints .items()}
chunks_midpoints = [x+start_pos for x in chunks_midpoints]
'''Combined into first interation through list of chunks above
#Calculate G vs. C and A vs. T that I think might give results like Figure 8
# of Morrill et al 2016 (PMID: 27026700)
#---------------------------------------------------------------------------
chunks_diffs = {} #key will be index of chunk in chunks. value will be tuple
# with each diff as an item
for indx,seq in enumerate(chunks):
GC_diff = calc_nt_diff("GC",str(seq)) # casting 'Sequence' object to
# string with `str(seq)`
AT_diff = calc_nt_diff("AT",str(seq)) # casting 'Sequence' object to
# string with `str(seq)`
chunks_diffs[indx] = (GC_diff,AT_diff)
'''
#Make the plot
#---------------------------------------------------------------------------
sns.set()
plt.figure(figsize=default_plt_image_size)
#indx = list(chunks_midpoints.values())
#data = list(chunks_diffs.values())
#indx and data were originally organized in dictionaries during development
# but for better compatibility with Python 2.7 lists were used so order
# maintained
#df = pd.DataFrame(data, indx, ["CvsG", "TvsA"])
df = pd.DataFrame(chunks_diffs, chunks_midpoints, ["G-C", "A-T"]) # went
# with what is used Morrill et al. figure for line labels even though
# `["CvsG", "TvsA"])` is more descriptive to what is being plotted.
ax = sns.lineplot(data=df)
ax.set_ylabel(yaxis_label, fontsize = 16);
ax.set_xlabel(xaxis_label_prefix+chr_, fontsize = 16);
ax.legend(fontsize= 12);
# Return plot figure object (meant for calling from Jupyter cell or IPython)
# or save to file.
#---------------------------------------------------------------------------
if return_fig:
sys.stderr.write("Plot figure object returned.")
return ax
else:
#save image; standard for when called from command line; however, will
# also be default for when called from Jupyter notebook / IPython unless
# called with `return_fig=True`
output_file_name = generate_output_file_name(sequence_file,
suffix_for_saving_plot)
if save_vg:
plt.savefig(output_file_name[:-4]+".svg",
orientation='landscape') # FOR VECTOR GRAPHICS; useful if merging
# into Adobe Illustrator. Based on
# https://neuroscience.telenczuk.pl/?p=331 ; I think ReportLab also
# outputs SVG?
sys.stderr.write("\nPlot image saved to: {}\n".format(
output_file_name[:-4]+".svg"))
else:
# save png
plt.savefig(output_file_name)
sys.stderr.write("\nPlot image saved to: {}\n".format(
output_file_name))
import seaborn as sns
import pandas as pd
import matplotlib.pyplot as plt
if __name__ == '__main__':
fmri = sns.load_dataset("fmri")
ax = sns.lineplot(x="timepoint", y="signal", data=fmri)
plt.show()
# print(df_orig['del_rate'])
print(
"median error rate/subs/ins/del original:", median_error
) #, 100*df_orig['subs_rate'].median(), 100*df_orig['ins_rate'].median(), 100*df_orig['del_rate'].median())
# g = sns.catplot(x="abundance_original", y="abundance_corrected", col="Depth", col_wrap=3,
# data=indata, kind="point", aspect=1)
# ax = sns.scatterplot(x="abundance_original", y="abundance_corrected", #col="Depth", # col_wrap=3, #hue="time",
# data=indata)
# g = sns.lmplot(x="transcript_cov", y="err_rate", hue="type", scatter= False, x_estimator = np.mean, #x_jitter = 0.1, #col="Depth",
# x_ci= 'sd', data=indata) #, col_wrap=2, height=3)
ax = sns.lineplot(x="transcript_cov",
y="err_rate",
hue="type",
ci='sd',
estimator='median',
data=indata)
ax.set_ylim(0, 12)
ax.set_xscale('log')
ax.set_xticks(
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100])
ax.set_ylabel("Error rate (%)")
ax.set_xlabel("Reads per transcript")
# g.set(ylim=(0,100))
# ax.set_ylabel("Abundance after correction")
# ax.set_xlabel("Abundance before correction")
# g.set_ylabels("Error rate (%)")
# g.set_xlabels("Reads per transcript")
def lineplot_per_interval(df,
y,
hue,
y_title="",
start_year=DEFAULT_MIN_YEAR,
save_to_file=True,
filetype="pdf",
legend_loc="upper right",
hue_order=None,
legend_title="Snapshot rank",
estimator=np.median):
fig, ax = plt.subplots()
if not y_title:
if hue:
y_title = hue.title()
else:
y_title = ""
# filter by start year
df = df[df.year_season >= str(start_year)]
set_plot_params()
if hue:
try:
fig = sns.lineplot(data=df,
y=y,
x=INTERVAL_COL_NAME,
hue=hue,
style=hue,
hue_order=hue_order,
estimator=estimator,
ci=95)
except:
markers = list(range(1, len(hue_order) + 1))
print(markers)
fig = sns.lineplot(data=df,
y=y,
x=INTERVAL_COL_NAME,
style=hue,
hue=hue,
dashes=False,
markers=markers,
hue_order=hue_order)
set_legend(ax, legend_title, legend_loc)
plt.setp(fig.get_legend().get_texts(),
fontsize='10') # for legend text
plt.setp(fig.get_legend().get_title(),
fontsize='10') # for legend text
else:
fig = sns.lineplot(data=df,
y=y,
x=INTERVAL_COL_NAME,
estimator=estimator)
set_x_ticks(df)
#fig.set(xlabel='Interval', ylabel=y_title)
ax.set_xlabel('Interval')
ax.set_ylabel(y_title, fontsize=10)
# fig.set_title(y_title + " per interval")
fig_filename = "lineplot_%s.%s" % ("_".join(
y_title.replace("%", "pct").replace("(", "").replace(
")", "").lower().split()), filetype)
if save_to_file:
save_figure(fig, fig_filename)
return fig
has_latest_model = True
predictions_latest, _ = dataset.predict(df, './models/latest', True)
print('Mean error latest : ' + str(dataset.mean_error(np.array(predictions_latest), y)))
success_rate_latest = dataset.get_trend_success_rate(predictions_latest, y, df)
print('Trend prediction success for latest : ' + str(success_rate_latest) + '%')
except:
has_latest_model = False
predictions_latest = None
pass
print('Mean error reference : ' + str(dataset.mean_error(np.array(predictions_reference), y)))
success_rate_reference = dataset.get_trend_success_rate(predictions_reference, y, df)
print ('Trend prediction success for reference : ' + str(success_rate_reference) + '%')
figure(num=None, figsize=(24, 10), dpi=80, facecolor='w', edgecolor='k')
ax = sns.lineplot(x=predictions_reference.index, y=y, label="Test Data", color='blue')
ax = sns.lineplot(x=predictions_reference.index, y=predictions_reference[0], label="Prediction reference", color='gray')
if has_latest_model:
ax = sns.lineplot(x=predictions_reference.index, y=predictions_latest[0], label="Prediction latest", color='red')
ax.set_title('Price', size = 14, fontweight='bold')
ax.set_xlabel("Hours", size = 14)
ax.set_ylabel("Cost", size = 14)
ax.set_xticklabels('', size=10)
plt.show()
homogeneity = {}
# use kmeans to loop over candidate number of clusters
# store inertia and homogeneity score in each iteration
for k in range(1, 5):
km = KMeans(n_clusters=k)
pred = km.fit_predict(x)
inertia[k] = km.inertia_
homogeneity[k] = homogeneity_score(y, pred)
# %%
ax = sns.lineplot(
x=list(inertia.keys()),
y=list(inertia.values()),
color="blue",
label="inertia",
legend=None,
)
ax.set_ylabel("inertia")
ax.twinx()
ax = sns.lineplot(
x=list(homogeneity.keys()),
y=list(homogeneity.values()),
color="red",
label="homogeneity",
legend=None,
)
ax.set_ylabel("homogeneity")
ax.figure.legend()
daily_data.dtypes
dfNW56Articles.reset_index(inplace=True, drop=True) #reset index
#merge
myArticles = pd.merge(dfNW56Articles, daily_data, on='date', sort=True)
myArticles.describe() #95 articles au final.
# Sauvegarde de myArticles en csv pourrait servir dans d'autres articles.
myArticles.to_csv("myArticles.csv", sep=";", index=False) #séparateur ;
##########################################################################
# Graphiques des événements et des publications des articles
# sur toute la période
##########################################################################
sns.set() #paramètres esthétiques ressemble à ggplot par défaut.
fig, ax = plt.subplots() #un seul plot
sns.lineplot(x='date', y='pageviews', data=daily_data, color='grey', alpha=0.2)
sns.scatterplot(x='date',
y='pageviews',
data=myOutliers,
color='red',
alpha=0.5)
sns.scatterplot(x='date',
y='pageviews',
data=myArticles,
color='blue',
marker="+")
fig.suptitle(str(len(myOutliers)) + " événements (ronds rouges) pour " +
str(len(myArticles)) + " articles (croix bleues) : ",
fontsize=14,
fontweight='bold')
ax.set(
train_df['year'] = train['date'].dt.year
train_df['month'] = train['date'].dt.month
train_df['day'] = train['date'].dt.dayofyear
train_df['weekday'] = train['date'].dt.weekday
train_df.head()
# --------------------------------------------------------------------------------------------------
# ---- Scomposizione Time Series
# per cominciare bisogna scomporre la serie per studiare:
# -- la stagionalità
# -- il trend
# -- i residui
# dato che abbiamo 5 anni di dati ci aspettiamo una stagionalità annuale o settimanale
sns.lineplot(x="date", y="sales",legend = 'full' , data=train_df)
# dal grafico sopra possiamo vedere un trend crescente ed una stagionalità annuale
# sembra che nei mesi centrali ci sia un picco nelle vendite
sns.lineplot(x="date", y="sales",legend = 'full' , data=train_df[:28])
# dal grafico sopra non vediamo chiaramente la presenza di una stagionalità settimanale
sns.boxplot(x="weekday", y="sales", data=train_df)
# da questo grafico (sopra) possiamo vedere come in media ci sia un lieve trend crescente
# nelle vendite all'interno della settimana
# Lunedì = 0 ---- Domenica = 6
# inoltre possiamo vedere che nei giorni infra-settimanali le vendite siano minori che nel weekend
# creo train_df
train_df = train_df.set_index('date')
train_df['sales'] = train_df['sales'].astype(float)
temp[i, j]
])
hv_df = pd.DataFrame(hv_progress_df,
columns=[
'Evaluations', 'Runs',
'Approaches', 'HV (Underlying)'
])
print(hv_df)
color_map = plt.cm.get_cmap('viridis')
color_map = color_map(np.linspace(0, 1, 5))
ax = sns.lineplot(x="Evaluations",
y="HV (Underlying)",
hue="Approaches",
style="Approaches",
markers=True,
dashes=False,
data=hv_df,
palette=color_map)
#box = ax.get_position()
handles, labels = ax.get_legend_handles_labels()
#ax.legend(handles=handles, labels=labels)
ax.legend(handles=handles[1:],
labels=labels[1:],
frameon=False,
loc='upper center',
bbox_to_anchor=(0.5, 1.15),
ncol=5)
#plt.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
#ax.set_title(prob + ', ' + str(obj) + ' objs, ' + sx)
# finding outliers of purchases sns.boxplot(cc_clean['purchases'], color='#fdc029') plt.show() # finding outliers of cash_advance sns.boxplot(cc_clean['cash_advance'], color='#fdc029') plt.show() # finding outliers of payments sns.boxplot(cc_clean['payments'], color='#fdc029') plt.show() # plotting balance and different purchase options plt.figure(figsize=(10,4)) sns.lineplot(cc_clean['balance'],cc_clean['purchases'],label="Purchases") sns.lineplot(cc_clean['balance'],cc_clean['oneoff_purchases'],label='Oneoff') sns.lineplot(cc_clean['balance'],cc_clean['installments_purchases'],label='Install') plt.show() sns.pairplot(cc_clean) """# Hierarchical clustering""" # using standardscaler to reduce the distance of each variable sc = StandardScaler() xs = sc.fit_transform(cc_clean) X = pd.DataFrame(xs, index=cc_clean.index, columns=cc_clean.columns) # hclust # going to do euclidean, cosine and cityblock distance
def do_plot(df, thr, min_epc, hue='dset', title='', legend_hue=False):
sns.set_theme()
scrtmarkers = {'optima': 'o', 'stop': 'X', 'optima/stop': 'D'}
plt.subplot(211)
if title:
plt.title(title)
gsc = sns.scatterplot(data=df[df['convergence'] != ''],
x='epoch',
y='err_prior',
hue=hue,
markers=scrtmarkers,
s=100,
style='convergence',
legend=True)
# get current handles and labels
current_handles, current_labels = plt.gca().get_legend_handles_labels()
# remove title
t_ = [(h, l) for h, l in zip(current_handles, current_labels)
if (('optima' in l) or ('stop' in l))]
current_handles = [t[0] for t in t_]
current_labels = [t[1] for t in t_]
conv_leg = plt.legend(current_handles,
current_labels,
bbox_to_anchor=(1.0, 1.0),
loc='upper right')
g = sns.lineplot(data=df,
x='epoch',
y='err_prior',
hue=hue,
legend=legend_hue)
if legend_hue:
# get handles of "hue"
n_hue = df[hue].unique().size
current_handles, current_labels = plt.gca().get_legend_handles_labels()
current_handles = current_handles[:n_hue]
current_labels = current_labels[:n_hue]
plt.legend(current_handles,
current_labels,
bbox_to_anchor=(0.75, 1.),
loc='upper right')
# We need this because the 2nd call to legend() erases the first #
g.add_artist(conv_leg)
g.set(xlabel=None)
plt.ylabel("Mean absolute error \n of priors")
plt.subplot(212)
g = sns.lineplot(data=df,
x='epoch',
y='var_pseudo_neg',
hue=hue,
legend=False)
plt.axhline(y=thr, color='r', linestyle='--')
plt.axvline(x=min_epc, color='g', linestyle='--')
g = sns.scatterplot(data=df[df['convergence'] != ''],
x='epoch',
y='var_pseudo_neg',
hue=hue,
legend=False,
markers=scrtmarkers,
s=100,
style='convergence')
plt.ylabel("variance of\n pseudo negatives")
return plt.gcf()
sns.barplot(data=four_most_popular, y='item_name', x='revenue')
plt.ylabel('')
plt.xticks()
# 5.) Load the sleepstudy data and read it's documentation.
# Use seaborn to create a line chart of all the individual subject's reaction times and a more prominant line showing the average change in reaction time.
sleep = data('sleepstudy')
data('sleep', show_doc=True)
sleep.info()
sleep.tail(20)
sleep.describe()
sns.set_style('darkgrid')
sns.set_context(font_scale=1,
rc={
"grid.linewidth": 1,
"axes.linewidth": 1,
"ytick.major.width": 1,
"xtick.major.width": 1,
"lines.linewidth": 1
})
palette = sns.color_palette("deep", 18)
sns.lineplot(x='Days',
y='Reaction',
hue='Subject',
data=sleep,
palette=palette)
sns.set_context(rc={"lines.linewidth": 4})
sns.lineplot(x='Days', y='Reaction', data=sleep, ci=None)
A = 1/(12 * K * (K * N))
B = (k + N - 1/2)/(K*(K+N))
func = lambda tau : -1/tau**2 + 2*A*tau + B
taus = []
for j in [10,500,100]:
tau_initial_guess = 1
tau_solution = fsolve(func, tau_initial_guess)
taus.append(tau_solution)
tau_solution = np.mean(taus)
points.append({"k":k,"alpha":tau_solution,"N":N,"K":K})
pfx = pd.DataFrame(points)
pfx['N'] = pfx.N.astype(str)
pfx.to_csv("simulation_data.tsv",sep="\t")
plt.savefig("simulation.pdf",dpi=300)
pfx = pd.read_csv("simulation_data.tsv",sep="\t")
sns.lineplot(pfx.K,pfx.alpha, label = "Simulated data")
def fit_func(x, a, b):
return a*np.power(x,2/3) + b
params = curve_fit(fit_func, pfx.K, pfx.alpha)
a, b = params[0]
X_space = pfx.K
Y_space = a * np.power(X_space,2/3) + b
plt.plot(X_space,Y_space, label = r"$"+str(np.round(a,2))+r" \cdot K^{\frac{2}{3}} "+str(np.round(b,2))+r"$")
plt.legend()
plt.ylabel(r"$\gamma$")
#plt.title(r"Fitting $\gamma$ estimate to the space of 250,000 networks")
plt.savefig("simulation_final.pdf",dpi=300)
def plot_linear(k,X,y,I_pca,I_spearman,I_mRMR,I_hierarchy,I_sklearn,I_permutation_oob,I_dropcol_oob):
df_linear = pd.DataFrame({'k':np.arange(1,k+1)
, 'mae_pca':mae_I_linear(X,y,sort_I(I_pca),k)
,'mae_spearman':mae_I_linear(X,y,sort_I(I_spearman),k)
,'mae_mRMR':mae_I(X,y,sort_I(I_mRMR),k)
,'mae_hierarchy':mae_I_linear(X,y,sort_I(I_hierarchy),k)
,'mae_sklearn':mae_I_linear(X,y,sort_I(I_sklearn),k)
,'mae_permutation_oob':mae_I_linear(X,y,sort_I(I_permutation_oob),k)
,'mae_dropcol_oob':mae_I_linear(X,y,sort_I(I_dropcol_oob),k)
})
sns.lineplot(df_linear['k'],df_linear['mae_pca'],marker='o', label='pca')
sns.lineplot(df_linear['k'],df_linear['mae_spearman'],marker='p', label='spearman')
sns.lineplot(df_linear['k'],df_linear['mae_mRMR'],marker='1', label='mRMR')
sns.lineplot(df_linear['k'],df_linear['mae_hierarchy'],marker='s', label='hierarchy')
sns.lineplot(df_linear['k'],df_linear['mae_sklearn'],marker='d', label='sklearn')
sns.lineplot(df_linear['k'],df_linear['mae_permutation_oob'],marker='*', label='permutation')
sns.lineplot(df_linear['k'],df_linear['mae_dropcol_oob'],marker=11, label='dropcol')
plt.ylabel('20% 5fold CV MAE ($)')
plt.title('linear model')
sep="\t",
names=["Ref", "Pos", "depth"])
sns.lmplot(x="POS", y="ALT_FREQ", data=rep_b, fit_reg=False)
fig, ax1 = plt.subplots(figsize=(15, 5))
color = "tab:green"
#isnv plot creation
ax1.set_ylabel('iSNVs', color="blue", fontsize=20)
ax1.tick_params(axis='y', labelcolor="blue")
ax1 = sns.regplot(x="POS", y="ALT_FREQ", data=rep_b, ax=ax1, palette='summer')
#specify we want to share same x axis
ax2 = ax1.twinx()
color = 'tab:red'
# lineplot creation for the depth
sns.lineplot(x='Pos', y='depth', data=japan1_depth, ax=ax2)
ax2.set_ylabel('Depth', color="red")
ax2.tick_params(axis='y', labelcolor="red")
plt.show()
fig, ax1 = plt.subplots()
ax1.set_xlabel('Position')
ax1.set_ylabel('iSNVs', color="blue")
ax1.tick_params(axis='y', labelcolor="blue")
rep_a.plot(x="POS",
y="ALT_FREQ",
label="1_peru",
ax=ax1,
kind="scatter",
color="blue",
# Sets figure parameters
plt.figure(figsize=(6,5))
plt.rcParams['axes.titlesize'] = 13
plt.rcParams['axes.labelsize'] = 13
plt.rcParams['xtick.labelsize'] = 13
plt.rcParams['ytick.labelsize'] = 13
plt.rcParams['legend.fontsize'] = 12
plt.xlabel('Number of tasks (T)', fontsize=13)
plt.ylabel('Makespan (a.u.)', fontsize=13)
plt.xticks(range(1000,10001,1000))
plt.xticks(rotation=15)
plt.ylim(0, 12000)
sns.lineplot(data=recursive_results[recursive_results.Resources == 10],
x='Tasks',
y='Makespan',
hue='Scheduler',
linewidth=2,
hue_order=order)
plt.savefig("s1-recursive-10.pdf", bbox_inches='tight')
# In[ ]:
# Sets figure parameters
plt.figure(figsize=(6,5))
plt.rcParams['axes.titlesize'] = 13
plt.rcParams['axes.labelsize'] = 13
plt.rcParams['xtick.labelsize'] = 13
plt.rcParams['ytick.labelsize'] = 13
fn, columns=["year_season", "policy_snapshot_url", "match_str"])
cache[identifier] = results_df
return results_df
def plot_term(df,
term,
regex=False,
case=False,
min_year=DEFAULT_MIN_YEAR,
save_figure=True):
plt.figure(figsize=(10, 5))
percentages = search_term(df, term, regex, case, min_year)
fig = sns.lineplot(x="interval", y="percentage", data=percentages)
fig.set_xticklabels(fig.get_xticklabels(), rotation=45, fontsize='small')
if not final:
title = "Query: %s\n(Min-year: %s, Regex: %s, Case sensitive: %s)" % (
term, min_year, regex, case)
fig.set_title(title)
plt.ylim(ymin=0)
if save_figure:
s_fig = fig.get_figure()
s_fig.savefig("figures/%s_%s_%s_%s.png" %
(term, min_year, regex, case),
bbox_inches='tight')
return fig
all_recalls[model] = mean_recall * 100
all_maps[model] = np.mean(maps) * 100
if APPROXIMATE:
print(all_recalls[model])
print(len(all_recalls[model]))
if len(all_precisions[model]) > 1000:
last_precision = all_precisions[model][-1]
last_recall = all_recalls[model][-1]
all_precisions[model] = all_precisions[model][0::50]
all_recalls[model] = all_recalls[model][0::50]
all_precisions[model] = np.append(all_precisions[model], last_precision)
all_recalls[model] = np.append(all_recalls[model], last_recall)
print(all_recalls[model])
print(len(all_recalls[model]))
sns.lineplot(all_precisions[model], all_recalls[model], label=str(model_print_name), linewidth=3)
plt.xlim([0.0, 100])
plt.ylim([0.0, 105])
plt.title("WHEN: Precision-recall plot")
plt.legend(loc="lower left")
sns.despine()
plt.savefig("./figures/roc_legend.png")
plt.xlim([0.0, 100])
plt.ylim([0.0, 105])
plt.title("WHEN: Precision-recall plot")
plt.legend(loc="lower left")
ax.get_legend().remove()
sns.despine()
plt.savefig("./figures/roc_when.png")
def plot(self, df_in, output_prefix, output_dir, dpi, force, show,
verbose):
df = df_in
# Set a filename if needed.
if self.filename is None:
self.filename = f"{self.plot_type}--{self.xkey}--{self.ykey}--{self.huekey}--{self.stylekey}"
# Use some seaborn deafault for fontsize etc.
sns.set_context(self.sns_context, rc={"lines.linewidth": 2.5})
# Set the general style.
sns.set_style(self.sns_style)
# Filter the data using pandas queries if required.
if self.df_query is not None and len(self.df_query):
df = df.query(self.df_query)
# If the df is empty, skip.
if df.shape[0] == 0:
print(f"Skipping plot {self.filename} - 0 rows of data")
return False
# Get the number of palette values required.
huecount = len(
df[self.huekey].unique()) if self.huekey is not None else 1
# Set palette.
palette = sns.color_palette(self.sns_palette, huecount)
sns.set_palette(palette)
# create a matplotlib figure and axis, for a single plot.
# Use constrained layout for better legend placement.
fig, ax = plt.subplots(constrained_layout=True)
# Set the size of the figure in inches
fig.set_size_inches(FIGSIZE_INCHES[0], FIGSIZE_INCHES[1])
# Generate labels / titles etc.
xlabel = f"{pretty_csv_key(self.xkey)}"
ylabel = f"{pretty_csv_key(self.ykey)}"
huelabel = f"{pretty_csv_key(self.huekey)}"
stylelabel = f"{pretty_csv_key(self.stylekey)}"
hs_label = f"{huelabel} x {stylelabel}" if huelabel != stylelabel else f"{huelabel}"
figtitle = f"{ylabel} vs {xlabel} (hs_label)"
# @todo - validate keys.
# Decide if using internal legend.
external_legend = self.legend_outside
filled_markers = ('o', 'v', '^', '<', '>', '8', 's', 'p', '*', 'h',
'H', 'D', 'd', 'P', 'X')
g = None
if self.plot_type == "lineplot":
# plot the data @todo - lineplot vs scatter?
g = sns.lineplot(
data=df,
x=self.xkey,
y=self.ykey,
hue=self.huekey,
style=self.stylekey,
markers=True,
dashes=True,
ax=ax,
# size=6,
legend=self.sns_legend,
palette=palette,
)
elif self.plot_type == "scatterplot":
g = sns.scatterplot(
data=df,
x=self.xkey,
y=self.ykey,
hue=self.huekey,
style=self.stylekey,
markers=True,
ax=ax,
# size=6,
legend=self.sns_legend,
palette=palette,
)
else:
raise Exception(f"Bad plot_type {self.plot_type}")
# Set a title
# @disabled for now.
# if len(figtitle):
# plt.title(figtitle)
# adjust x axis if required.
ax.set(xlabel=xlabel)
if self.logx:
ax.set(xscale="log")
if self.minx is not None:
ax.set_xlim(left=self.minx)
if self.maxx is not None:
ax.set_xlim(right=self.maxx)
# adjust y axis if required.
ax.set(ylabel=ylabel)
if self.logy:
ax.set(yscale="log")
if self.miny is not None:
ax.set_ylim(bottom=self.miny)
if self.maxy is not None:
ax.set_ylim(top=self.maxy)
# Disable scientific notation on axes
ax.ticklabel_format(useOffset=False, style='plain')
# If there is reason to have a legend, do some extra processing.
if ax.get_legend() is not None:
legend = ax.get_legend()
loc = None
bbox_to_anchor = None
if external_legend:
# Set legend placement if not internal.
loc = "upper left"
# @todo - y offset should be LEGNED_BORDER_PAD trasnformed from font units to bbox.
bbox_to_anchor = (1, 1 - self.legend_y_offset)
# Get the handles and labels for the legend
handles, labels = ax.get_legend_handles_labels()
# Iterate the labels in the legend, looking for the huekey or stylekey as values
# If either were found, replace with the pretty version
found_count = 0
for i, label in enumerate(labels):
if label == self.huekey:
labels[i] = huelabel
found_count += 1
elif label == self.stylekey:
labels[i] = stylelabel
found_count += 1
# If neither were found, set a legend title.
if found_count == 0:
# add an invisble patch with the appropriate label, like how seaborn does if multiple values are provided.
handles.insert(
0,
mpatches.Rectangle((0, 0),
1,
1,
fill=False,
edgecolor='none',
visible=False,
label=hs_label))
labels.insert(0, hs_label)
pass
ax.legend(handles=handles,
labels=labels,
loc=loc,
bbox_to_anchor=bbox_to_anchor,
borderaxespad=LEGEND_BORDER_PAD)
# if an output directory is provided, save the figure to disk.
if output_dir is not None:
output_dir = pathlib.Path(output_dir)
# Prefix the filename with the experiment prefix.
output_filename = f"{output_prefix}--{self.filename}"
# Get the path for output
output_filepath = output_dir / output_filename
# If the file does not exist, or force is true write the otuput file, otherwise error.
if not output_filepath.exists() or force:
try:
if verbose:
print(f"writing figure to {output_filepath}")
fig.savefig(output_filepath, dpi=dpi, bbox_inches='tight')
except Exception as e:
print(
f"Error: could not write to {output_filepath} with exception {e}"
)
return False
else:
print(
f"Error: {output_filepath} already exists. Specify a different `-o/--output-dir` or use `-f/--force`"
)
return False
# If not outputting, or if the show flag was set, show the plot.
if show: # or output_dir is None:
plt.show()
return True
from statsmodels.sandbox.predict_functional import predict_functional
# In[104]:
values = {"hist": 0, "tumorsize": 50, "accinsitu": 0, "lymphinv": 0}
# In[105]:
pr, cb, fv = predict_functional(result,
"age",
values=values,
ci_method="simultaneous")
# In[106]:
ax = sns.lineplot(fv, pr, lw=4)
ax.fill_between(fv, cb[:, 0], cb[:, 1], color='grey', alpha=0.4)
ax.set_xlabel("age")
ax.set_ylabel("Re-excision")
ax.set_title('Fitted Model: Log-odd probability of Age by Re-excision')
#This plot of fitted log-odds visualizes the effect of age on reexcision for
#hist=0, tumorsize=23, accinsitu=0 and lumphinv=0 by the glm fitted model
#Slight negative correlation of age and RE are visible in this plot
#For the specific described variables
# In[ ]:
# In[100]:
bins=range(
int(company.Low.min()) - 1,
int(company.Low.max()) + 1),
normed=True,
rwidth=1)
plt.xticks(range(int(company.Low.min()) - 1, int(company.Low.max()) + 1))
plt.xlabel("Lowest Prize for each day")
plt.ylabel("Probability")
plt.title("Company: " + name.split(".")[0], fontsize=20)
plt.show()
# In[18]:
for name, i, company in zip(companies_9, range(9), data):
plt.figure(figsize=(15, 5))
sns.lineplot(data=data[i], x='Date', y="Open", label="Opening Value")
sns.lineplot(data=data[i], x='Date', y="Close", label="Closing Value")
sns.lineplot(data=data[i],
x='Date',
y="High",
label="Highest Value Per Day")
sns.lineplot(data=data[i], x='Date', y="Low", label="Lowest Value Per Day")
plt.legend()
#plt.xticks([])
plt.xlabel("Time")
plt.ylabel("Stock Prices")
plt.title("Compnay: " + name.split(".")[0])
plt.show()
# In[23]:
#Print dataframe
print("")
print("")
print("=" * 72)
print(" " * 3,
"Average time taken to run each sorting algorithm in milliseconds")
print("=" * 72)
print(df)
print("=" * 72)
totalEndTime = time.time()
# Print time taked to execute program
totalTimeElapsed = totalEndTime - totalStartTime
print("")
print("Program took ", round(totalTimeElapsed, 3), " seconds to run")
# Testing functions
#Save data in csv to plot graphs in seperate program quicker
#df.to_csv('sortingDataframe.csv')
# Plot data
plot = sns.lineplot(data=df, markers=True,
dashes=False) # plot using seaborn
plot.set(xlabel='Array Size',
ylabel='Time taken (milliseconds)') # label axis
#plt.yscale("log") # Uncomment to graph log scale results for larger number of inputs
plt.show() # Show plot
plt.style.use("ggplot")
# plt.rcParams.update({'legend.fontsize': 14})
p = sns.color_palette()
sns.set_palette([p[0], p[1]])
f, axes = plt.subplots(1, 1, figsize=(width, height))
# sns.lineplot(x='samples', y='success_rate', hue='algo', ax=axes[0], data=sr_timesteps)
# axes[0].set_xlabel('samples')
# axes[0].set_ylabel('success_rate')
# axes[0].get_legend().remove()
# sns.lineplot(x='samples', y='eval', hue='algo', ax=axes[1], data=eval_timesteps)
# axes[1].set_xlabel('samples')
# axes[1].set_ylabel('')
# axes[1].get_legend().remove()
sns.lineplot(x='iterations',
y='len_mean',
hue='algo',
ax=axes,
data=len_mean_iteration)
axes.set_xlabel('iterations')
axes.set_ylabel('episode length')
axes.get_legend().remove()
handles, labels = axes.get_legend_handles_labels()
# if mode == 'train':
# sns.lineplot(x='samples', y='success_rate', hue='algo', data=sr_timesteps)
# axes.set_xlabel('samples')
# elif mode == 'hard':
# sns.lineplot(x='samples', y='eval', hue='algo', data=eval_timesteps)
# axes.set_xlabel('samples')
# elif mode == 'iteration':
# sns.lineplot(x='iterations', y='eval', hue='algo', ax=axes, data=eval_iteration)
# axes.set_xlabel('iterations')
"bleu": [0, 0],
"data_ord": [0, 0],
"data": ["nejm.0", "nejm.0"],
"direction": ["zh2en", "en2zh"],
"train": ["de novo", "de novo"]
})
bleu = pd.concat([zeros, bleu])
ord2num = {0: 0, 1: 1, 2: 2, 3: 3, 4: 4, 5: 5, 6: 5.46}
bleu["x"] = bleu["data_ord"].apply(lambda x: ord2num[x])
plt.ion()
fig, ax = plt.subplots(1, 1)
g = sns.lineplot(x="x",
y="bleu",
hue="direction",
data=bleu,
legend="brief",
style="train",
markers=["o", "o"],
dashes=[(2, 1), ""])
fig.set_size_inches(5, 4)
fig.tight_layout()
g.legend_.texts[0].set_position((-40, 0))
g.legend_.texts[0].set_text("Direction")
g.legend_.texts[3].set_position((-40, 0))
g.legend_.texts[3].set_text("Training")
ax.set_xlabel("In-Domain Sentence Pairs")
ax.set_ylabel("1-ref BLEU")
ax.set_xticks([0, 1, 2, 3, 4, 5, 5.46])
ax.set_xticklabels(["0", "4000", "8000", "16000", "32000", "64000", ""])
ax.legend()
plt.savefig(f"{out_dir}/bleu.pdf")
kde_kws={'linewidth': 3},
label=race)
# Plot formatting
plt.legend(prop={'size': 16}, title='Race')
plt.title('Victim Race and Median Income (2017 - 2018)')
plt.xlabel('Median Income')
plt.ylabel('Density')
# line chart for victim race and income
#sns.lineplot(x="timepoint", y="signal", data=fmri)
sns.lineplot(x='Victim Descent',
y='Median Income',
data=la,
markers=True,
dashes=False)
#plot formatting
#plt.legend(prop={'size': 16}, title = 'Race')
plt.title('Victim Descent and Median Income (2017 - 2018)')
# line chart for victim race and income
#sns.lineplot(x="timepoint", y="signal", data=fmri)
sns.lineplot(x='Victim Age', y='Median Income', data=la)
#plot formatting
#plt.legend(prop={'size': 16}, title = 'Race')
"""
Timeseries plots with error bands
=================================
_thumb: .5, .45
"""
import seaborn as sns
sns.set(style="darkgrid")
# Load an example dataset with long-form data
fmri = sns.load_dataset("fmri")
# Plot the responses for different events and regions
sns.lineplot(x="timepoint", y="signal",
hue="region", style="event",
data=fmri)
df_acc = pd.DataFrame(index=ary_epochs)
for ind in range(len(lst_evnt_trn)):
# add data to pandas loss data frame
df_loss['trn_loss' + lst_names[ind]] = lst_trn_lss[ind]
df_loss['val_loss' + lst_names[ind]] = lst_val_lss[ind]
# add data to pandas loss data frame
df_acc['trn_acc' + lst_names[ind]] = lst_trn_acc[ind]
df_acc['val_acc' + lst_names[ind]] = lst_val_acc[ind]
# plot losses
fig, ax = plt.subplots()
fig.set_size_inches(17.5, 12.5)
sns.lineplot(data=df_loss,
palette=lst_colors,
dashes=lst_dashes,
linewidth=2.5)
plt.xlabel("Number of Epochs")
plt.ylabel("Loss")
fig.savefig(
"/media/sf_D_DRIVE/Unet/presentation/results/plots/loss_model_weighted.svg"
)
fig.savefig(
"/media/sf_D_DRIVE/Unet/presentation/results/plots/loss_model_weighted.png"
)
# plot accuracies
fig, ax = plt.subplots()
fig.set_size_inches(17.5, 12.5)
sns.lineplot(data=df_acc, palette=lst_colors, dashes=lst_dashes, linewidth=2.5)
plt.xlabel("Number of Epochs")
def plot(self, data=None, **fargs):
if data is None:
data = self.df
data = data.melt("timeStamp", var_name="cols", value_name="vals")
sns.lineplot(data=data, **fargs)
pass
test_scorecard_array = np.asarray(test_scorecard)
test_performance.append(test_scorecard_array.sum() /
test_scorecard_array.size)
print("testing performance = ",
test_scorecard_array.sum() / test_scorecard_array.size)
pass
# plot performance
sns.set(rc={'figure.figsize': (15, 10)})
performance_x = np.vstack((train_performance, test_performance)).T
performance_x_df = pd.DataFrame(performance_x).reset_index()
df = performance_x_df.melt('index', var_name='train/test', value_name='vals')
df['train/test'] = df['train/test'].map({0: 'train', 1: 'test'})
performance_x_plot = sns.lineplot(x="index",
y="vals",
hue='train/test',
data=df)
performance_x_plot.set(xlabel='EPOCHS', ylabel='PERFORMANCE')
plt.savefig('performance_x.png')
# confusion matrix
a_x = pd.Series(np.asarray(actual))
p_x = pd.Series(np.asarray(predicted))
confusion_matrix_x = pd.crosstab(a_x, p_x)
confusion_matrix_x
# for experiment 2...
# shuffle and sample quarter of the training data
np.random.shuffle(train_data)
quarter_train_data = train_data[:15000]
sns.get_dataset_names()
# ## lineplot
# In[61]:
fmri = sns.load_dataset("fmri")
# In[20]:
fmri.head()
# In[33]:
ax = sns.lineplot(x="timepoint", y="signal", err_style="band", data=fmri)
# In[28]:
ax = sns.lineplot(x="timepoint", y="signal", err_style="bars", data=fmri)
# In[63]:
ax = sns.lineplot(x="timepoint", y="signal", ci=95, color="m", data=fmri)
ax = sns.lineplot(x="timepoint", y="signal", ci=68, color="b", data=fmri)
# In[35]:
ax = sns.lineplot(x="timepoint", y="signal", ci='sd', color="m", data=fmri)
# In[37]:
import seaborn as sns
data = pd.read_csv("Processed-text.csv",
low_memory=False,
index_col=[2],
parse_dates=[2]).sort_index()
data.drop(columns=['Unnamed: 0', 'body'], inplace=True, axis=1)
#Sales Distribution basis on comment
monthly_order_count = data.resample('M').count()
#orders_monthly.dropna(how='any',inplace=True)
monthly_order_count = pd.DataFrame(monthly_order_count)
#plt.plot_date(x=orders_monthly.index,y=orders_monthly['rating'],fmt='o')
monthly_order_count.rename(columns={'rating': 'Sales'}, inplace=True)
ax = sns.lineplot(x=monthly_order_count.index,
y=monthly_order_count['Sales'],
data=monthly_order_count,
markers=True)
#Overall sentiment over time period
orders_monthly = data.resample('M').median()
orders_monthly = pd.DataFrame(orders_monthly)
ax = sns.lineplot(x=orders_monthly.index,
y=orders_monthly['rating'],
data=orders_monthly,
markers=True)
import plotly.graph_objs as go
import plotly.offline as py
actual_chart = go.Scatter(x=orders_monthly.index,
y=orders_monthly['Sales'],
df.loc[:, 'dh'] = df.loc[:, 'h'].diff() / pd.Timedelta(dt).total_seconds()
step = 1
for t in tides_rep_with_slr.index[1:]:
t_min_1 = t - pd.Timedelta(dt)
df.loc[t, 'z'] = calc_z(df.at[t_min_1, 'z'], df.at[t_min_1, 'dz'], 0, 0)
df.loc[t, 'C0'] = calc_c0(
df.at[t, 'h'], df.at[t, 'dh'], df.at[t, 'z'], A, SSC)
df.loc[t, 'C'] = calc_c(df.at[t, 'C0'], df.at[t, 'h'], df.at[t_min_1, 'h'],
df.at[t, 'dh'], df.at[t_min_1, 'C'], df.at[t, 'z'], ws, dt_sec)
df.loc[t, 'dz'] = calc_dz(df.at[t, 'C'], ws, rho, dt_sec)
print('Completed step {0} of {1}.'.format(step, len(tides_rep_with_slr)))
step = step + 1
# %%
f, ax = plt.subplots(figsize=(5,6))
sns.set_style("whitegrid")
plt_data = df.shift(freq=pd.Timedelta(days=1326))
plt_data = plt_data[::10]
sns.lineplot(data=plt_data.h, alpha = 0.25)
sns.lineplot(data=plt_data.z, color = 'black')
plt.title("Tidal Heights and Land Surface")
plt.xlabel("Year")
plt.ylabel("Height above Mean Water Level (m)")
plt.xlim(start + pd.Timedelta(days=1326), rep_end + pd.Timedelta(days=1326))
plt.show()
#%%
#tides_rep.tail()
