data_loading.py

import os
import pandas as pd

from .errors import compute_errors

def load_costs(building_strategy_costs, floor_strategy_costs):
    building_df = pd.read_csv(building_strategy_costs)
    floor_df = pd.read_csv(floor_strategy_costs)
    building_df['Strategy'] = 'Building'
    floor_df['Strategy'] = 'Floor'

    df = pd.concat([building_df, floor_df])
    return df[['Strategy', 'N', 'N_CLUSTERS', 'COMP_TIME']].groupby(['Strategy', 'N', 'N_CLUSTERS']).mean().reset_index()


def load_raw_positioning_errors(DIR_PATH):
    model_results = {}
    for clustering_features in os.listdir(DIR_PATH):
        features_dir = os.path.join(DIR_PATH, clustering_features)
        if not os.path.isdir(features_dir):
            continue
        for results_csv in os.listdir(features_dir):
            if not results_csv.endswith('csv'):
                continue
    
            components = results_csv.split('_')
            model = components[0]
            strategy = components[-2]
            raw_results = pd.read_csv(os.path.join(DIR_PATH, clustering_features, results_csv))
            model_results.setdefault(clustering_features, {}).setdefault(strategy, {})[model] = compute_errors(raw_results)
    return model_results