四半期の単一時系列データのグラフ化

単一の時系列データ（頻度は四半期）の特徴を示す
いつもの1行目。

import pandas as pd
from datetime import datetime as dt
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import scipy.optimize
import statsmodels.api as sm

データ読み込み。

df = pd.read_csv('data/best_forecast.csv', index_col=0)
df.index=pd.to_datetime(df.index)

これはデータの分布を示すfunction。

def prob_prot(np_array, ax):
    #use this func as:axes[1]=prob_prot(df_ret.dropna().values, axes[1])
    from statistics import mean, stdev
    mu = mean(np_array)  # mean of distribution
    sigma = stdev(np_array)  # standard deviation of distribution

    num_bins = 10

    #fig, ax = plt.subplots()

    # the histogram of the data
    n, bins, patches = ax.hist(np_array, num_bins, density=1)

    # add a 'best fit' line
    y = ((1 / (np.sqrt(2 * np.pi) * sigma)) *
         np.exp(-0.5 * (1 / sigma * (bins - mu))**2))
    ax.plot(bins, y, '--')
    ax.set_title(r'Histogram: $\mu=$'+ str(round(mu,3)) +r', $\sigma=$' + str(round(sigma,3)))
    return ax

季節性をグラフで示すfunction。

def seaz_prot(df_diff, ax):
    from statistics import mean, stdev
    ret_list = []
    ret_list_index = []
    ret_mean_list= []
    for i in range(0,4):
        j=df_diff.tail(4).index.month[i]
        ret=df_diff[df_diff.index.month==j].values
        ret_list.extend(ret)
        ret_list_index.extend([j for k in range(0,len(ret))])
        ret_mean_list.extend([np.nanmean(ret)])
    df = pd.DataFrame({'series': ret_list}, index=ret_list_index)
    df_mean = pd.DataFrame({'series': ret_mean_list}, index=df_diff.tail(4).index.month.tolist())
    df_mean.sort_index(inplace=True)
    ax.scatter(df.index, df['series'])
    ax.set_xticks(df_mean.index)
    ax.set_xticklabels =df_mean.index
    ax.plot(df_mean.index, df_mean['series'],color=diff_color)
    for a,b in zip(df_mean.index,df_mean['series']): 
        ax.text(a, b, str(round(b,3)))
    ax.set_title('Seasonality')
    return ax

さまざまなデータの特徴を1枚のグラフで示す

df_diff=(df[:dt.today()]['IS_OPER_INC']-df[:dt.today()]['BEST_OPP'])/df[:dt.today()]['BEST_OPP']
df_ret=df_diff
df_ret[~((df_ret < 0.3) & (df_ret > -0.3))]=np.nan
plt.rcParams["font.size"] = 10
fig, ax = plt.subplots(2, 3, figsize=(20, 8))
axes = ax.flatten()
diff_color = "#800000"

axes[0].plot(df_ret)
axes[0].set_title("Raw data")
axes[0].grid(which='major',color='gray',linestyle='-')

axes[1]=prob_prot(df_ret.dropna().values, axes[1])
axes[1].grid(which='major',color='gray',linestyle='-')

fig = sm.graphics.tsa.plot_acf(df_diff.dropna(), lags=5, ax=axes[2])
axes[2].grid(which='major',color='gray',linestyle='-')

diff1_data=df_diff-df_diff.shift()
axes[3]=seaz_prot(df_diff, axes[3])
axes[3].grid(which='major',color='gray',linestyle='-')

res = sm.tsa.seasonal_decompose(df_diff.dropna().values, freq=4)
df_res = pd.DataFrame({'Trend': res.trend, 'Residual': res.resid}, index=df_diff.dropna().index)
axes[4].plot(df_res['Trend'])
axes[5].plot(df_res['Residual'])

axes[4].grid(which='major',color='gray',linestyle='-')
axes[4].set_title('Trend')
axes[5].grid(which='major',color='gray',linestyle='-')
axes[5].set_title('Residual (=Raw data - Seasonality - Trend)')
plt.show()