Clean features across participants.

Explore the best linear regression feature.

exploration/ex_ml_pipeline.py

@@ -21,12 +21,14 @@ import os
 import sys
 
 import numpy as np
+import matplotlib.pyplot as plt
 import pandas as pd
 import seaborn as sns
 import yaml
 from pyprojroot import here
 from sklearn import linear_model
 from sklearn.model_selection import LeaveOneGroupOut, cross_val_score
+from sklearn.metrics import mean_squared_error, r2_score
 from sklearn.impute import SimpleImputer
 
 nb_dir = os.path.split(os.getcwd())[0]
@@ -378,3 +380,76 @@ sns.heatmap(features_labels[feature_columns].isna(), cbar=False)
 # ```
 
 # %%
+features_rapids_cleaned = pd.read_csv(here("rapids/data/processed/features/all_participants/all_sensor_features_cleaned_rapids.csv"), parse_dates=["local_segment_start_datetime", "local_segment_end_datetime"])
+features_rapids_cleaned = features_rapids_cleaned.assign(date_lj=lambda x: x.local_segment_start_datetime.dt.date)
+features_rapids_cleaned["participant_id"] = features_rapids_cleaned["pid"].str.extract("(\d+)")
+features_rapids_cleaned["participant_id"] = pd.to_numeric(features_rapids_cleaned["participant_id"])
+features_rapids_cleaned.set_index(["participant_id", "date_lj"], inplace=True)
+
+# %%
+features_cleaned_labels = features_rapids_cleaned.join(labels_read, how="inner").reset_index()
+feature_clean_columns = features_cleaned_labels.columns[6:-3]
+
+# %%
+print(feature_columns.shape)
+print(feature_clean_columns.shape)
+
+# %%
+sns.set(rc={"figure.figsize":(16, 8)})
+sns.heatmap(features_cleaned_labels[feature_clean_columns].isna(), cbar=False)
+
+# %%
+lin_reg_rapids_clean = linear_model.LinearRegression()
+logo = LeaveOneGroupOut()
+logo.get_n_splits(
+    features_cleaned_labels[feature_clean_columns],
+    features_cleaned_labels[label_column],
+    groups=features_cleaned_labels[group_column],
+)
+
+# %%
+features_clean_imputed = imputer.fit_transform(features_cleaned_labels[feature_clean_columns])
+
+# %%
+cross_val_score(
+    lin_reg_rapids_clean,
+    X=features_clean_imputed,
+    y=features_cleaned_labels[label_column],
+    groups=features_cleaned_labels[group_column],
+    cv=logo,
+    n_jobs=-1,
+    scoring="r2",
+)
+
+# %%
+lin_reg_full = linear_model.LinearRegression()
+lin_reg_full.fit(features_clean_imputed,features_cleaned_labels[label_column])
+
+# %%
+NA_pred = lin_reg_full.predict(features_clean_imputed)
+
+# %%
+# The coefficients
+print("Coefficients: \n", lin_reg_full.coef_)
+# The mean squared error
+print("Mean squared error: %.2f" % mean_squared_error(features_cleaned_labels[label_column], NA_pred))
+# The coefficient of determination: 1 is perfect prediction
+print("Coefficient of determination: %.2f" % r2_score(features_cleaned_labels[label_column], NA_pred))
+
+# %%
+feature_clean_columns[np.argmax(lin_reg_full.coef_)]
+
+# %% [markdown]
+# Ratio between stationary time and total location sensed time. A lat/long coordinate pair is labeled as stationary if its speed (distance/time) to the next coordinate pair is less than 1km/hr. A higher value represents a more stationary routine.
+
+# %%
+plt.scatter(features_clean_imputed[:,np.argmax(lin_reg_full.coef_)], features_cleaned_labels[label_column], color="black")
+plt.scatter(features_clean_imputed[:,np.argmax(lin_reg_full.coef_)], NA_pred, color="red", linewidth=3)
+
+plt.xticks()
+plt.yticks()
+
+fig = plt.gcf()
+fig.set_size_inches(18.5, 10.5)
+
+plt.show()