def get_classes(df):
classes = []
feat_list = parse.get_features_list(df)
for feature in df:
# Skip the Index column and the features that do not contain exclusively numeric values
if feature not in feat_list:
continue
classes.append(feature)
return classes
def my_scatter_plot(df):
fig = go.Figure()
feat_list = parse.get_features_list(df)
for feature in df:
# Skip the Index column and the features that do not contain exclusively numeric values
if feature not in feat_list:
continue
fig.add_trace(
go.Scatter(y=df[feature],
name=feature,
opacity=0.8,
mode='markers'))
fig.update_layout(title="Normalized grades for each class",
xaxis_title="Index",
yaxis_title="Normalized grades")
fig.show()
def plot_hist(df):
i = 0
j = 0
fig, axs = plt.subplots(4, 4)
feat_list = parse.get_features_list(df)
for feature in df:
if feature not in feat_list:
continue
for house in cst.houses:
axs[i, j].hist(get_feature_per_house(df, house, feature),
bins=25,
alpha=0.5,
color=cst.houses_colors[house])
axs[i, j].set_title(feature)
j += 1
if j == 4:
i += 1
j = 0
plt.tight_layout()
plt.show()
def my_histogram(df):
std = []
feat_name = []
n_feat = 0
feat_list = parse.get_features_list(df)
for feature in df:
if feature not in feat_list:
continue
count = describe.get_count(df[feature])
std.append(
describe.get_std(count, describe.get_mean(count, df[feature]),
df[feature]))
feat_name.append(feature)
n_feat += 1
plt.bar(range(n_feat), std, color=cst.colors)
plt.xticks(range(n_feat), feat_name, rotation=-45, fontsize=6, ha="left")
plt.title(
"Standard deviation between the student's grades for each feature \n(less std means that the student's grades are homogeneous)"
)
plt.show()
def test_describe(df, describe, print_describe=True):
# Values calculated with system/numpy functions go to control_values dict
# It will allow us to check if our functions return correct values
feat_list = parse.get_features_list(df)
control_values = df.describe().loc[:, feat_list[0]:]
errors = 0
for feature in df:
# Skip the features that do not contain exclusively numeric values
if feature not in feat_list:
continue
# For each significative difference between results from our functions and results from system/numpy functions, we output a warning
for value in control_values[feature].keys():
if not np.isclose(describe[feature][value], control_values[feature][value]):
errors += 1
warn_diff(feature, value, describe[feature][value], control_values[feature][value])
if print_describe is True:
print(tabulate(describe, headers="keys", tablefmt="fancy_grid", floatfmt=".6f"))
print(tabulate(control_values, headers="keys", tablefmt="fancy_grid", floatfmt=".6f"))
return errors