analysis.py

import pandas as pd
import numpy as np
import plotly.express as px
import plotly.figure_factory as ff


def results_to_df(results: dict, metric_name: str):
    metric_scores = []
    for topic, results_dict in results.items():
        for metric_name_cur, metric_value in results_dict.items():
            if metric_name == metric_name_cur:
                metric_scores.append(metric_value)  
    return pd.DataFrame({metric_name: metric_scores})


def create_boxplot_1df(results: dict, metric_name: str):
    df = results_to_df(results, metric_name)
    fig = px.box(df, y=metric_name)
    return fig


def create_boxplot_2df(results1, results2, metric_name):
    df1 = results_to_df(results1, metric_name)
    df2 = results_to_df(results2, metric_name)
    df2["Run"] = "Run 2"
    df1["Run"] = "Run 1"
    df = pd.concat([df1, df2])

    # Create distplot with custom bin_size
    fig = px.histogram(df, x=metric_name, color="Run", marginal="box", hover_data=df.columns)
    return fig


def create_boxplot_diff(results1, results2, metric_name):
    df1 = results_to_df(results1, metric_name)
    df2 = results_to_df(results2, metric_name)
    diff = df1[metric_name] - df2[metric_name]

    x_axis = f"Difference in {metric_name} from 1 to 2"
    fig = px.histogram(pd.DataFrame({x_axis: diff}), x=x_axis, marginal="box")
    return fig