def update_figure(selected_dropdown1, selected_dropdown2, val, fr):
dropdown = {
'Coursera': cou_n,
'edX': edx_n,
'Khan Academy': kha_n,
'Pluralsight': plu_n,
'Skillshare': ski_n,
'Udacity': uda_n,
'Udemy': ude_n,
'Skype': sky_n,
'Zoom': zoo_n,
'TED Talks': ted_n
}
df = dropdown[selected_dropdown1]
if val == 1:
df1 = df[df['username'] == selected_dropdown1]
elif val == 2:
df1 = df[df['username'] != selected_dropdown1]
else:
df1 = df
test = data_prep(df1)
test_i = seasonal_decompose(test[selected_dropdown2],
model='additive',
period=fr)
t1 = pd.DataFrame(test_i.trend)
t1.index = t1.index.to_timestamp().to_pydatetime()
figure = {
'data': [{
'x': t1.index,
'y': t1['trend'],
'range_x': [t1.index.min(), t1.index.max()],
'type': 'scatter',
'mode': 'marker',
'opacity': 0.7
}],
'layout': {
'title':
'Trend analysis of 2019-20 period for {}'.format(
selected_dropdown2),
'xaxis': {
'title': 'Timeline'
}
}
}
return figure
RUN_CREATE_FEATURE_DATASET = False
RUN_CREATE_ROC_CAUSAL_DATASET = False
RUN_EXPERIMENTS = False #RUN_CREATE_FEATURE_DATASET also needs to be true
'''Latents Dimension for Deconfounder Algorithm (DA)'''
k_mf_ = [40]
k_pca_ = [40]
k_ac_ = [10]
'''Loading dataset'''
filename = "data\\tcga_train_gexpression_cgc_7k.txt" #_2
filename_gamma = "results\\gamma.txt"
#Running Factor Analysis Models + Predictive Check + outcome model in all patients
if RUN_ALL:
data = pd.read_csv(filename, sep=';')
#data = data.iloc[0:500, 0:100]
train, j, v, y01, abr, colnames = fc.data_prep(data)
#j: rows, v: columns, y01: initial label, abr: cancer type, colnames: genes names
'''
1) Run factor model;
2) Do predictive check;
3) If pass on predictive ckeck, run outcome model
4) Save results and predictions for ROC curve
'''
df_gamma = pd.read_csv(filename_gamma, sep=';')
gamma = []
cil = []
cip = []
id2 = []
if RUN_MF:
for k_mf in k_mf_:
import functions
#improvements: change variable names to be more intuitive
chdir("/home/cree/Downloads/br_econ/")
count = 0
mun, state = {}, {}
mun = functions.import_data("mun")
state = functions.import_data("state")
chdir("/home/cree/workspace/econometrics/")
# municipal dataframes
mun_renda = functions.data_prep(mun.get('mun/Renda_municipios - Renda familiar - per capita - media.csv'))
mun_ensino = functions.data_prep(mun.get('mun/mun_media_anos_de_estudos_25_anos_+_todos.csv'))
mun_saude = mun.get('mun/mun_Mortalidade infantil (por mil nascidos vivos)_1970-2000.csv')
mun_populacao_censo = functions.data_prep(mun.get('mun/populacao_municipal.csv'))
mun_renda_2, mun_ensino_2 = mun_renda.ix[3:].copy(),mun_ensino.ix[3:].copy()
mun_renda, mun_ensino = mun_renda.ix[:3],mun_ensino.ix[:3]
#state dataframes
state_renda = functions.data_prep(state.get('state/estado_Renda domiciliar per capita - media_1976 - 2014.csv'))
state_ensino = functions.data_prep(state.get('state/Anos de estudo - media - pessoas 25 anos e mais - 1981 - 2014.csv'))
state_ensino_mulheres = functions.data_prep(state.get('state/Anos de estudo - media - pessoas 25 anos e mais - mulheres 1981 -2014.csv'))
state_populacao_anual = functions.data_prep(state.get('state/populcao_estado_1980-2014.csv'))
state_populacao_censo = state.get('state/populacao_residente_estado_1970-2000.csv')
#calcular mortalidade infantil por estado
df9 = state_populacao_censo[['Sigla', 'Código', 'Estado']].copy()
def update_figure(selected_dropdown1, val):
dropdown = {
'Coursera': cou_n,
'edX': edx_n,
'Khan Academy': kha_n,
'Pluralsight': plu_n,
'Skillshare': ski_n,
'Udacity': uda_n,
'Udemy': ude_n,
'Skype': sky_n,
'Zoom': zoo_n,
'TED Talks': ted_n
}
df = dropdown[selected_dropdown1]
if val == 1:
df1 = df[df['username'] == selected_dropdown1]
var = selected_dropdown1 + ' handle'
elif val == 2:
df1 = df[df['username'] != selected_dropdown1]
var = 'Users'
else:
df1 = df
var = 'Both (Platform and user)'
test = data_prep(df1)
test.index = test.index.to_timestamp().to_pydatetime()
trace1 = go.Bar(x=test.index,
y=test['tweet_counter'] - test['label'],
name='Non Covid',
hovertext=test['tweet_counter'] - test['label'])
trace2 = go.Bar(x=test.index,
y=test['label'],
name='Covid',
hovertext=test['label'])
figure1 = go.Figure(
data=[trace1, trace2],
layout=go.Layout(
barmode='stack',
title=
'Analysis for Covid and Non-Covid related tweets from {} on a weekly basis '
.format(var),
# paper_bgcolor='rgba(48, 48, 48, 1)',
plot_bgcolor='white'))
df2 = df1[df1['label'] == 1]
wc = [len(text) for text in df2['text']]
time = [tim for tim in df2['timestamp']]
trace3 = go.Scatter(x=time,
y=wc,
mode='markers',
marker={
'color': '#ff471a',
'size': wc,
'sizemode': 'area',
'sizeref': 2. * max(wc) / (40.**2),
'sizemin': 4
})
figure2 = go.Figure(data=[trace3],
layout=go.Layout(
plot_bgcolor='white',
title='Covid related tweets on timeline'))
return figure1, figure2
'Pluralsight': plu_n,
'Skillshare': ski_n,
'Udacity': uda_n,
'Udemy': ude_n,
'Skype': sky_n,
'Zoom': zoo_n,
'TED Talks': ted_n
}
server = flask.Flask(__name__)
app = dash.Dash(__name__, server=server)
data = list(dict_main.keys()) # keys for the dataset values
channels = dict_main[data[0]] # the_data_sets
channels = data_prep(channels)
app.layout = html.Div([
html.H1('COVID Edu-Tech', style={'textAlign': 'center'}),
html.Div([
html.Div([
html.Label([
"Select Platform",
dcc.Dropdown(id='data-dropdown',
options=[{
'label': label,
'value': label
} for label in data],
value=list(dict_main.keys())[0],
multi=False,
searchable=False,