-
Notifications
You must be signed in to change notification settings - Fork 2
/
preprocess_sorted.py
142 lines (125 loc) · 5.13 KB
/
preprocess_sorted.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
print("Imports...", end="")
import datetime
import csv
import numpy as np
import json
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
import sys
sns.set()
pd.set_option('display.max_columns', None)
print("done")
with open('params.json', 'r') as f:
params = json.load(f)
typeable = []
for _, v in params.get('fingers-to-keys').items():
typeable.extend(v)
typeable = list(set(typeable))
print("Preprocessing data from sorted.log")
# -------------------------------------------
# Read in and clean up the original dataframe
# -------------------------------------------
# Read in data from tsv. Use python engine because it's more feature complete
df = pd.read_csv("sorted.log", "\t", names=['micros', 'sensor', 'value'], quoting=csv.QUOTE_NONE)
print("Dataframe read in successfully")
# Convert micros since epoch to datetime,
df['datetime'] = pd.to_datetime(df['micros'], unit='us')
# Remove the irrelevant micros column
del df['micros']
# Drop all the sensors we're not using
df = df[df['sensor'].isin(params.get('sensors-in-use'))]
# Rename the sensors from indices to names
replacers = {int(k): v for k, v in params.get('index-to-name').items()}
df[['sensor']] = df[['sensor']].replace(to_replace=replacers)
# -------------------------------
# Calculate the Sensors Dataframe
# -------------------------------
print("Calculating the Sensors Dataframe")
# Store the key presses separately from the sensor value readings
sensors = df.loc[df['sensor'] != 'keyboard']
# The sensor's values are all numerical data readings
sensors.loc[:, 'value'] = pd.to_numeric(sensors.loc[:, 'value'])
# Remove outliers
l, h = sensors.value.quantile([0.0001, 0.9999])
sensors = sensors[sensors.value.between(l, h)]
# resample
period_size = 10 if len(sys.argv) != 2 else int(sys.argv[1])
period_sizetd = pd.Timedelta(period_size, unit="ms")
# The data might be spread out over days or weeks which makes resampling for
# every 10ms very RAM intensive. Rather split up the data by date and resample
# on the parts
sensors['date'] = sensors.datetime.dt.date
gb = sensors.groupby('date')
sensor_dfs = [gb.get_group(group) for group in gb.groups]
print("Resampling the sensors dataframe")
resampled_sensors = []
for sensor_df in sensor_dfs:
resampled_sensors.append(sensor_df.groupby('sensor').resample(period_sizetd, on='datetime').median().reset_index())
sensors = pd.concat(resampled_sensors)
# Remove any time points which have no sensor readings.
sensors = sensors.dropna()
now_as_iso = datetime.datetime.isoformat(datetime.datetime.now())[:-16]
sensors_path = f"sensors-{now_as_iso}-{period_size}ms.pkl"
sensors.to_pickle(f"sensors-{now_as_iso}-{period_size}ms.pkl")
print(f"Saved sensors as {sensors_path}")
del sensors
# ----------------------------
# Calculate the keys Dataframe
# ----------------------------
print("Calculating the keys dataframe")
# Get a df with the data
keys = df.loc[(df['sensor'] == 'keyboard') & (df['value'].isin(typeable))]
# Lower-case all the values because we don't know if a shift key has been pressed
def fancy_lower(x):
""" lowercase _everything_, as though the shift key doesn't exist"""
if x.isalpha():
return x.lower()
if x in params.get('lower-dict').keys():
return params.get('lower-dict').get(x)
return x
print("Lower casing all the key presses")
keys.value = keys.value.apply(fancy_lower)
# The data might be spread out over days or weeks which makes resampling for
# every 10ms very RAM intensive. Rather split up the data by date and resample
# on the parts
keys['date'] = keys.datetime.dt.date
gb = keys.groupby('date')
key_dfs = [gb.get_group(group) for group in gb.groups]
print("Re sampling the keys dataframe")
resampled_keys = []
for key_df in key_dfs:
resampled_keys.append(key_df.set_index('datetime').resample(period_sizetd).first().reset_index().dropna())
keys = pd.concat(resampled_keys)
del keys['date']
print("Saving the dataframes")
keys_path = f"keys-{now_as_iso}-{period_size}ms.pkl"
keys.to_pickle(keys_path)
print(f"Saved keys as {keys_path}")
# print(f"Total Readings from keyboard: "
# + str(df[(df['sensor'] == 'keyboard')].value.notna().sum()))
# print(f"Total Readings from sensors: ")
# print(df[df['sensor'] != 'keyboard'].groupby('sensor').count().value)
# print("Making chart...")
# if input("Make chart?"):
# sns.lineplot(data=sensors, x='datetime', y='value', hue='sensor', lw=0.5)
# def plt_keys(row):
# plt.text(
# x=row['datetime'],
# y=sensors['value'].max(),
# s=f"'{row['value']}'",
# alpha=0.5
# )
# plt.axvline(x=row['datetime'], alpha=0.5, c='black', linewidth=1)
#
# keys.apply(plt_keys, axis='columns')
# plt.show()
# Xy = sensors
# # TODO: this is super slow ):
# print("Starting slow transformation of the data's datetimes")
# Xy['key'] = Xy.datetime.apply(lambda x: (keys.loc[keys.datetime==x, 'value'].unique()[0] if (keys.datetime==x).any() else np.NaN))
# freq_keys = Xy['key'].value_counts().index.to_list()[:10]
# Faceted boxenplot
# g = sns.FacetGrid(Xy[Xy.key.isin(freq_keys)], col='sensor', col_wrap=2)
# g.map(sns.boxplot, 'value', 'key', palette='Set3')
# plt.show()