File size: 5,811 Bytes
8cf4695
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
import pandas as pd
import numpy as np
import math

import matplotlib.pyplot as plt
import seaborn as sns

from statsmodels.graphics.tsaplots import plot_pacf
from statsmodels.tsa.seasonal import MSTL, seasonal_decompose
from statsmodels.tsa.tsatools import freq_to_period

import statsmodels.api as sm

from scipy import stats
from sklearn.preprocessing import MinMaxScaler


class Ts_Analytics():
    def __init__(self):
        self.log_transformed = False
        self.scaler = MinMaxScaler()
        pass

    def analyse(
            self,
            ts_df: pd.DataFrame,
            auto_correlations={}):
        ''' 
        ts_df: timeseries dataframe, will assume the datetime column is the index and time encoded
        auto_correlations: dictionary to input customised auto correlations
        '''
        self.ts_df = ts_df.copy()

        self.ts_df['datetime'] = pd.to_datetime(self.ts_df['datetime'])
        self.ts_df.set_index('datetime', inplace=True)

        self.ar = auto_correlations

        self.__infer_frequency()

        # Annual maps to 1, quarterly maps to 4, monthly to 12, weekly to 52.
        # Using statsmodel's freq_to_period function
        self.period = freq_to_period(self.freq)
        pass

    def set_ar(self, col, ar):
        ''' 
        Set the auto correlation
        '''
        self.ar[col] = ar

    def set_period(self, period):
        self.period = period

    def create_target_lag_columns(self):
        print('create_target_lag_columns')

        def create_lag_dfs(col, n):
            print('create_lag_dfs', col, n)
            for i in range(n):
                self.ts_df[f'{col}_t-{i+1}'] = self.ts_df[col].shift(-(i+1))

        for col, n in self.ar.items():
            create_lag_dfs(col, n)

        print('drop all null values')
        self.ts_df.ffill(inplace=True)

    def log_transform(self):
        self.log_transformed = True
        self.ts_df = np.log2(self.ts_df)

    def exp_transform(self):
        self.log_transformed = False
        self.ts_df = np.exp(self.ts_df)

    def train_multiple_regression(self):
        print('train_multiple_regression')

        x_cols = self.ts_df.columns.tolist()
        x_cols.remove('y')

        _X = self.ts_df[x_cols]
        y = self.ts_df['y']

        X = sm.add_constant(_X)

        # ----------------------------------------------------------------------- #
        # Train an additional model with standardized data, to get the Beta value #
        # ----------------------------------------------------------------------- #
        std_ts_df = pd.DataFrame(self.scaler.fit_transform(
            self.ts_df), columns=self.ts_df.columns)

        std_X = sm.add_constant(std_ts_df[x_cols])
        std_y = std_ts_df['y']

        self.multiple_regression = sm.OLS(y, X).fit()

        coef = self.multiple_regression.params

        self.multiple_regression_formula = f'{coef[0]} + {" + ".join([f"{c} * {round(n, 3)}" for c, n in zip(x_cols, coef[1:])]) }'

        self.std_multiple_regression = sm.OLS(std_y, std_X).fit()
        beta = self.std_multiple_regression.params

        self.multiple_regression_beta = pd.DataFrame(
            np.array(beta[1:]) ** 2, index=x_cols, columns=['Beta (influence on "y")'])
        self.multiple_regression_beta['Beta (influence on "y")'] = self.multiple_regression_beta['Beta (influence on "y")'].round(
            3)

        return self.multiple_regression.summary()

    # ===== #
    # Plots #
    # ===== #

    def plot_correlation(self):
        # Generate a mask for the upper triangle
        corr = self.ts_df.corr(numeric_only=True)
        mask = np.triu(np.ones_like(corr, dtype=bool))
        fig, ax = plt.subplots(figsize=(8, 8))

        sns.heatmap(
            corr,
            mask=mask,
            square=True,
            annot=True,
            cmap='coolwarm',
            linewidths=.5,
            cbar_kws={"shrink": .5},
            ax=ax)

        return fig

    def plot_target_pacf(self):
        fig, ax = plt.subplots(figsize=(12, 4))
        plot_pacf(self.ts_df['y'], ax=ax)
        fig.tight_layout()
        return fig

    def plot_distributions(self):
        plot_col = min(math.ceil(math.sqrt(self.ts_df.shape[1])), 5)
        plot_row = math.ceil(self.ts_df.shape[1] / plot_col)

        fig, axs = plt.subplots(plot_row, plot_col)

        for idx, col in enumerate(self.ts_df.columns):

            axs_x = math.floor(idx/plot_col)
            axs_y = idx - axs_x * plot_col

            # sns.distplot(self.ts_df[col], ax=axs[axs_x, axs_y])
            sns.histplot(self.ts_df[col], ax=axs[axs_x, axs_y], kde=True)

        fig.tight_layout()

        return fig

    def plot_target_seasonality(self):

        if isinstance(self.period, list):
            seasonal = MSTL(
                self.ts_df['y'], periods=self.period).fit()
        else:
            seasonal = seasonal_decompose(self.ts_df['y'], period=self.period)
        return seasonal

    def plot_beta(self):
        fig, ax = plt.subplots(figsize=(6, 4))

        beta_plot = sns.barplot(
            self.multiple_regression_beta['Beta (influence on "y")'], gap=2, ax=ax)
        beta_plot.set_xticklabels(beta_plot.get_xticklabels(), rotation=45)
        ax.bar_label(ax.containers[-1], fmt='%.2f', label_type='center')
        return fig

    def __infer_frequency(self):
        # Attempt to get the frequency from the provided datetime column
        freq = pd.infer_freq(self.ts_df.index)
        if freq is not None:
            self.freq = freq

        # Always make sure the frequency is not None
        if self.freq is None:
            raise ValueError(
                'Unable inference freq from datetime column, please make timeseries interval consistent or provide customized frequency.')