def asssistant():
values = request.json
data = {'n': len(values.values()[0]), 'k': len(values.keys())}
# Paired?
values_clean = clean_missing_values(request.json)
data['paired'] = True if data['n'] == len(
values_clean.values()[0]) else False
# Normality?
alpha = 0.1
values_clean = clean_missing_values(request.json, delete_row=False)
_, p_values = map(list,
zip(*[st.shapiro(v) for v in values_clean.values()]))
data['normality'] = int(
reduce(lambda x, y: x and y,
[p_value < alpha for p_value in p_values]))
# Homocedasticity?
_, p_value = st.levene(*values_clean.values())
data['homocedasticity'] = int(p_value < alpha)
data.update(evaluate_test(data))
return data
def mannwhitneyu(alpha=0.05):
values = clean_missing_values(request.json, delete_row=False)
statistic, p_value = st.mannwhitneyu(values.values()[0],
values.values()[1],
use_continuity="false")
result = int(p_value * 2 < alpha)
return {"result": result, "statistic": statistic, "p_value": p_value * 2}
def ttest_ind(alpha=0.05):
values = clean_missing_values(request.json)
statistic, p_value = st.ttest_ind(values.values()[0], values.values()[1])
result = int(p_value < alpha)
return {
"statistic": statistic.tolist(),
"p_value": p_value,
"result": result
}
def quade(alpha=0.05, post_hoc="bonferroni_dunn_test", control=None):
values = clean_missing_values(request.json)
statistic, p_value, rankings, ranking_cmp = npt.quade_test(
*values.values())
rankings, names = map(
list, zip(*sorted(zip(rankings, values.keys()), key=lambda t: t[0])))
ranks = {key: ranking_cmp[i] for i, key in enumerate(values.keys())}
if post_hoc.split('_')[-1] == "test":
comparisons, z_values, _, adj_p_values = getattr(npt,
post_hoc)(ranks,
control)
else:
comparisons, z_values, _, adj_p_values = getattr(npt, post_hoc)(ranks)
return statistic, p_value, rankings, names, comparisons, z_values, adj_p_values
def anova(alpha=0.05):
values = clean_missing_values(request.json)
statistic, p_value, pivots = pt.anova_within_test(*values.values())
pivots_cmp = {key: pivots[i] for i, key in enumerate(values.keys())}
comparisons, t_values, _, adj_p_values = pt.bonferroni_test(
pivots_cmp, len(values.values()[0]))
return {
"anova": {
"statistic": statistic,
"p_value": p_value,
"result": np.asscalar(p_value < alpha)
},
"post_hoc": {
"comparisons": comparisons,
"statistic": t_values,
"p_value": adj_p_values,
"result":
[int(adj_p_value < alpha) for adj_p_value in adj_p_values]
}
}
def wilcoxon(alpha=0.05):
values = clean_missing_values(request.json)
statistic, p_value = st.wilcoxon(values.values()[0], values.values()[1])
result = int(p_value < alpha)
return {"result": result, "statistic": statistic, "p_value": p_value}
def binomialsign(alpha=0.05):
values = clean_missing_values(request.json)
statistic, p_value = npt.binomial_sign_test(values.values()[0],
values.values()[1])
result = int(p_value < alpha)
return {"result": result, "statistic": statistic, "p_value": p_value}
def levene(alpha=0.05):
values = clean_missing_values(request.json, delete_row=False)
statistic, p_value = st.levene(*values.values())
result = int(p_value < alpha)
return {"statistic": statistic, "p_value": p_value, "result": result}
def agostino(alpha=0.05):
values = clean_missing_values(request.json, delete_row=False)
statistics, p_values = map(
list, zip(*[st.normaltest(v) for v in values.values()]))
result = [int(p_value < alpha) for p_value in p_values]
return {"statistic": statistics, "p_value": p_values, "result": result}
