forked from plexagon/lucius-ltv
-
Notifications
You must be signed in to change notification settings - Fork 0
/
plots.py
68 lines (56 loc) · 2.88 KB
/
plots.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
import seaborn as sns
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
import pandas as pd
import numpy as np
import arviz as az
def plot_ltv(empirical_ltv, inference_data=None, hdi_prob=.95, extra_label_text='', ax=None):
if ax is None:
fig, ax = plt.subplots(figsize=(20, 10))
if inference_data:
az.plot_hdi(x=np.arange(52), hdi_prob=hdi_prob, smooth=False, y=inference_data['posterior']['ltv'], ax=ax)
curve_m = np.median(np.median(inference_data['posterior']['ltv'], axis=0), axis=0)
ax.plot(curve_m, 'k', linestyle='dashed', alpha=0.5,
label=f'Median ltv: {curve_m[len(curve_m) - 1].round(2)}')
if 'true_ltv' in inference_data['posterior'].keys():
curve_m = np.median(np.median(inference_data['posterior']['true_ltv'], axis=0), axis=0)
ax.plot(curve_m, 'k', alpha=0.5,
label=f'True ltv: {curve_m[len(curve_m) - 1].round(2)}')
ax.plot(empirical_ltv, 'o',
label=f'{extra_label_text}Empirical @{len(empirical_ltv)} periods: {empirical_ltv[len(empirical_ltv) - 1].round(2)}')
return ax
def plot_conversion_rate(inference_data, hdi_prob=.95, extra_label_text='', ax=None):
if ax is None:
fig, ax = plt.subplots(figsize=(20, 10))
conversion_rate_by_cohort = inference_data['posterior']['conversion_rate_by_cohort']
az.plot_hdi(x=np.arange(conversion_rate_by_cohort.shape[-1]), hdi_prob=hdi_prob, smooth=False,
y=conversion_rate_by_cohort, ax=ax)
curve_m = np.median(np.median(inference_data['posterior']['conversion_rate_by_cohort'], axis=0), axis=0)
ax.plot(curve_m, 'k-', alpha=0.3,
label=f'{extra_label_text}Median Conversion Rate: {np.median(curve_m).round(2)}')
ax.plot(curve_m, 'ko', alpha=0.8,
label=f'{extra_label_text}Median Conversion Rate: {np.median(curve_m).round(2)}')
def plot_cohort_matrix_retention(cohort_matrix, title=''):
cohort_size = cohort_matrix.max(axis=1)
retention_matrix = cohort_matrix.divide(cohort_size, axis=0)
with sns.axes_style("white"):
fig, ax = plt.subplots(1, 2, figsize=(20, 10), sharey='all', gridspec_kw={'width_ratios': [1, 11]})
sns.heatmap(retention_matrix.iloc[:, :-1],
mask=retention_matrix.iloc[:, :-1].isnull(),
annot=True,
fmt='.0%',
cmap='RdYlGn',
ax=ax[1])
ax[1].set_title(title, fontsize=12)
ax[1].set(xlabel='# of periods',
ylabel='')
# cohort size
cohort_size_df = pd.DataFrame(cohort_size).rename(columns={0: 'cohort_size'})
white_cmap = mcolors.ListedColormap(['white'])
sns.heatmap(cohort_size_df,
annot=True,
cbar=False,
fmt='g',
cmap=white_cmap,
ax=ax[0])
fig.tight_layout()