93 lines
5.9 KiB
Python
93 lines
5.9 KiB
Python
import pandas as pd
|
|
import numpy as np
|
|
import datetime
|
|
import itertools
|
|
from datetime import datetime, timedelta, time
|
|
from features_utils import splitOvernightEpisodes, splitMultiSegmentEpisodes
|
|
|
|
def getEpisodeDurationFeatures(screen_deltas, episode, metrics):
|
|
screen_deltas_episode = screen_deltas[screen_deltas["episode"] == episode]
|
|
duration_helper = pd.DataFrame()
|
|
if "sumduration" in metrics:
|
|
duration_helper = pd.concat([duration_helper, screen_deltas_episode.groupby(["local_start_date"]).sum()[["time_diff"]].rename(columns = {"time_diff": "screen_" + day_segment + "_sumduration" + episode})], axis = 1)
|
|
if "maxduration" in metrics:
|
|
duration_helper = pd.concat([duration_helper, screen_deltas_episode.groupby(["local_start_date"]).max()[["time_diff"]].rename(columns = {"time_diff": "screen_" + day_segment + "_maxduration" + episode})], axis = 1)
|
|
if "minduration" in metrics:
|
|
duration_helper = pd.concat([duration_helper, screen_deltas_episode.groupby(["local_start_date"]).min()[["time_diff"]].rename(columns = {"time_diff": "screen_" + day_segment + "_minduration" + episode})], axis = 1)
|
|
if "avgduration" in metrics:
|
|
duration_helper = pd.concat([duration_helper, screen_deltas_episode.groupby(["local_start_date"]).mean()[["time_diff"]].rename(columns = {"time_diff":"screen_" + day_segment + "_avgduration" + episode})], axis = 1)
|
|
if "stdduration" in metrics:
|
|
duration_helper = pd.concat([duration_helper, screen_deltas_episode.groupby(["local_start_date"]).std()[["time_diff"]].rename(columns = {"time_diff":"screen_" + day_segment + "_stdduration" + episode})], axis = 1)
|
|
|
|
duration_helper = duration_helper.fillna(0)
|
|
return duration_helper
|
|
|
|
def getEventFeatures(screen_data, metrics_events, phone_sensed_bins, bin_size):
|
|
if screen_data.empty:
|
|
return pd.DataFrame(columns=["screen_" + day_segment + "_" + x for x in metrics_events])
|
|
# get count_helper
|
|
screen_status = screen_data.groupby(["local_date", "screen_status"]).count()[["timestamp"]].reset_index()
|
|
count_on = screen_status[screen_status["screen_status"] == 0].set_index("local_date")[["timestamp"]].rename(columns = {"timestamp": "count_on"})
|
|
count_off = screen_status[screen_status["screen_status"] == 1].set_index("local_date")[["timestamp"]].rename(columns = {"timestamp": "count_off"})
|
|
count_lock = screen_status[screen_status["screen_status"] == 2].set_index("local_date")[["timestamp"]].rename(columns = {"timestamp": "count_lock"})
|
|
count_unlock = screen_status[screen_status["screen_status"] == 3].set_index("local_date")[["timestamp"]].rename(columns = {"timestamp": "count_unlock"})
|
|
|
|
count_helper = pd.concat([count_on, count_off, count_lock, count_unlock], axis = 1)
|
|
count_helper = count_helper.fillna(0).astype(np.int64)
|
|
|
|
# get unlocks per minute
|
|
for date, row in count_helper.iterrows():
|
|
sensed_minutes = phone_sensed_bins.loc[date, :].sum() * bin_size
|
|
unlocks_per_minute = min(row["count_lock"], row["count_unlock"]) / (1 if sensed_minutes == 0 else sensed_minutes)
|
|
count_helper.loc[date, "unlocks_per_minute"] = unlocks_per_minute
|
|
|
|
event_features = pd.DataFrame()
|
|
if "counton" in metrics_events:
|
|
event_features["screen_" + day_segment + "_counton"] = count_helper[["count_on", "count_off"]].min(axis=1)
|
|
if "countunlock" in metrics_events:
|
|
event_features["screen_" + day_segment + "_countunlock"] = count_helper[["count_lock", "count_unlock"]].min(axis=1)
|
|
if "unlocksperminute" in metrics_events:
|
|
event_features["screen_" + day_segment + "_unlocksperminute"] = count_helper["unlocks_per_minute"]
|
|
|
|
return event_features
|
|
|
|
screen_data = pd.read_csv(snakemake.input["screen_events"], parse_dates=["local_date_time", "local_date"])
|
|
screen_deltas = pd.read_csv(snakemake.input["screen_deltas"], parse_dates=["local_start_date_time", "local_end_date_time", "local_start_date", "local_end_date"])
|
|
phone_sensed_bins = pd.read_csv(snakemake.input["phone_sensed_bins"], parse_dates=["local_date"], index_col="local_date")
|
|
phone_sensed_bins[phone_sensed_bins > 0] = 1
|
|
|
|
day_segment = snakemake.params["day_segment"]
|
|
metrics_events = snakemake.params["metrics_events"]
|
|
metrics_deltas = snakemake.params["metrics_deltas"]
|
|
episodes = snakemake.params["episodes"]
|
|
bin_size = snakemake.params["bin_size"]
|
|
|
|
metrics_deltas_name = ["".join(metric) for metric in itertools.product(metrics_deltas, episodes)]
|
|
|
|
if screen_data.empty:
|
|
screen_features = pd.DataFrame(columns=["local_date"]+["screen_" + day_segment + "_" + x for x in metrics_events + metrics_deltas_name])
|
|
else:
|
|
# drop consecutive duplicates of screen_status keeping the last one
|
|
screen_data = screen_data.loc[(screen_data[["screen_status"]].shift(-1) != screen_data[["screen_status"]]).any(axis=1)].reset_index(drop=True)
|
|
|
|
# preprocess day_segment and episodes
|
|
screen_deltas = splitOvernightEpisodes(screen_deltas, [], ["episode"])
|
|
if day_segment != "daily":
|
|
screen_data = screen_data[screen_data["local_day_segment"] == day_segment]
|
|
screen_deltas = splitMultiSegmentEpisodes(screen_deltas, day_segment, [])
|
|
screen_deltas.set_index(["local_start_date"],inplace=True)
|
|
|
|
# extract features for events and episodes
|
|
event_features = getEventFeatures(screen_data, metrics_events, phone_sensed_bins, bin_size)
|
|
|
|
if screen_deltas.empty:
|
|
duration_features = pd.DataFrame(columns=["screen_" + day_segment + "_" + x for x in metrics_deltas_name])
|
|
else:
|
|
duration_features = pd.DataFrame()
|
|
for episode in episodes:
|
|
duration_features = pd.concat([duration_features, getEpisodeDurationFeatures(screen_deltas, episode, metrics_deltas)], axis=1)
|
|
|
|
screen_features = pd.concat([event_features, duration_features], axis = 1).fillna(0)
|
|
screen_features = screen_features.rename_axis("local_date").reset_index()
|
|
|
|
screen_features.to_csv(snakemake.output[0], index=False) |