jcheong0428

!pip install -q statsmodels
import numpy as np
import pandas as pd
import statsmodels.api as sm
import statsmodels.formula.api as smf
data = sm.datasets.get_rdataset('dietox', 'geepack').data
md = smf.mixedlm("Weight ~ Time", data, groups=data["Pig"], re_formula="~Time")
mdf = md.fit(method=["lbfgs"])
print(mdf.summary())

jcheong0428

jcheong0428

jcheong0428

jcheong0428

f,axes = plt.subplots(2,3, figsize=(10,5))
for faces_ix, face_id in enumerate(df_clean.face_id.unique()[::-1]):
  df_clean.query(f'face_id=={face_id}').plot.scatter(x='gaze_angle_x', y='gaze_angle_y', ax=axes[0][faces_ix])
  axes[0][faces_ix].scatter(0,0, marker='x', color = 'k') # draw origin.
  axes[0][faces_ix].set(xlim=[-2,2], ylim=[-2,2], title=f'Gaze movement of face_id=={face_id}')
  df_clean.query(f'face_id=={face_id}')[['gaze_angle_x', 'gaze_angle_y']].plot(ax=axes[1][faces_ix])
  axes[1][faces_ix].set(ylim=[-1.5,1.5])
plt.tight_layout()
plt.show()

jcheong0428

# Threshold data by 80%
df_clean = df[df.confidence>=.80]
# Plot all Action Unit time series. 
au_regex_pat = re.compile(r'^AU[0-9]+_r$')
au_columns = df.columns[df.columns.str.contains(au_regex_pat)]
print("List of AU columns:", au_columns)
f,axes = plt.subplots(6, 3, figsize=(10,12), sharex=True, sharey=True)
axes = axes.flatten()
for au_ix, au_col in enumerate(au_columns):
    sns.lineplot(x='frame', y=au_col, hue='face_id', data=df_clean, ax=axes[au_ix])

jcheong0428

import re
x_regex_pat = re.compile(r'^x_[0-9]+$')
y_regex_pat = re.compile(r'^y_[0-9]+$')
x_locs = df.columns[df.columns.str.contains(x_regex_pat)]
y_locs = df.columns[df.columns.str.contains(y_regex_pat)]
palette = sns.color_palette("RdBu_r", 4)

avg_face_df = pd.DataFrame({'x_locs':df[x_locs].mean(axis=1), 'y_locs':df[y_locs].mean(axis=1), 'face_id': df.face_id})
ax = sns.scatterplot(x='x_locs', y='y_locs', hue = 'face_id', data=avg_face_df, marker="+", palette=palette)
ax.set(xlim=[0, 1920], ylim=[1080,0], title="Before thresholding");

jcheong0428

df.groupby('face_id').mean()['confidence']

jcheong0428

# See how many unique faces there are
print("Number of unique faces: ", len(df.face_id.unique()), "\nList of face_id's: ", df.face_id.unique())

jcheong0428

%matplotlib inline
%config InlineBackend.figure_format = 'retina'
import pandas as pd, seaborn as sns
sns.set_style('white')
import matplotlib.pyplot as plt 

# Load data
df = pd.read_csv('processed/video.csv')
# Remove empty spaces in column names.
df.columns = [col.replace(" ", "") for col in df.columns]