poi-analysis/tests/test_baseline_predictions.py

import unittest

import pandas as pd
from pyprojroot import here


from poi_analysis.baseline import BaselineRecommendationAlgorithm
from poi_analysis.clean_data import clean_poi_raw, clean_user_preferences


# CONFIG
DROP_VENUES = ['sophie café and pub tatra house',
               'waaberi restaurant & takeaway',
               'croft ales',
               'taste of morocco',
               'farro bakery']


class TestBaselineRecommendationAlgorithm(unittest.TestCase):

    def setUp(self):
        self.df_poi = clean_poi_raw(pd.read_excel(
            here() / "data/mottli_bristol_poi_curated/Bristol Places Datasheet v2.xlsx"
        ), exclude_names=DROP_VENUES)
        self.df_user = clean_user_preferences(pd.read_excel(
            here() / "data/mottli_bristol_poi_curated/Mottli - User preferences.xlsx"
        ), df_poi=self.df_poi)

    def test_score_all_users(self):
        bra = BaselineRecommendationAlgorithm(self.df_poi)
        preds = bra.score_all_users(self.df_user)

        # assert shape as expected
        self.assertTrue(preds.shape == (150, 12))
        # assert real values exist roughly in the region we expect
        self.assertTrue(preds.select_dtypes('float').mean().mean() < 0.8)
        self.assertTrue(preds.select_dtypes('float').mean().mean() > 0.4)

    def tearDown(self):
        pass