rapids/src/models/select_targets.py

import pandas as pd


cleaned_sensor_features = pd.read_csv(snakemake.input["cleaned_sensor_features"])

column_names = cleaned_sensor_features.columns
esm_names_index = column_names.str.startswith("phone_esm_straw")
# Find all columns coming from phone_esm, since these are not features for our purposes and we will drop them.
esm_names = column_names[esm_names_index]

target_variable_name = esm_names.str.contains(snakemake.params["target_variable"])
if all(~target_variable_name):
    raise ValueError("The requested target (", snakemake.params["target_variable"],  ")cannot be found in the dataset.",
                     "Please check the names of phone_esm_ columns in all_sensor_features_cleaned_rapids.csv")

model_input = cleaned_sensor_features.drop(esm_names, axis=1)
model_input["target"] = cleaned_sensor_features[esm_names[target_variable_name]]
# We will only keep one column related to phone_esm and that will be our target variable.
# Add it back to the very and of the data frame and rename it to target.

model_input.to_csv(snakemake.output[0], index=False)