【運営担当者様】気象データ(nyc_weather_2017_2019)の使用方法についてのご質問

天気データの使用方法

  • 予測対象の時間に対して、何時間前のデータまで使えるのかをご教授ください
  • 該当時刻のギリギリ直前までの使用が可能でしょうか?
import pandas as pd
pd.options.display.max_columns = 100
pd.options.display.max_rows = 999
pd.options.display.float_format = '{:.6f}'.format
import numpy as np

import datetime

import warnings
warnings.filterwarnings('ignore')

%matplotlib inline
# %matplotlib notebook
import matplotlib
import matplotlib.pyplot as plt
from matplotlib import cm
from google.colab import drive
drive.mount('/content/drive')
Mounted at /content/drive
path = "/content/drive/MyDrive/tmp/7_probspace/3_taxi/"

weather = pd.read_csv(path+'data/nyc_weather_2017_2019.csv', parse_dates=["DATE"]).rename(columns={'DATE':'ds'})
train = pd.read_csv(path+'data/train_data.csv', parse_dates=["tpep_pickup_datetime"]).rename(columns={'tpep_pickup_datetime':'ds'})
train[train["ds"]>="2019-11-30"].head(5)
ds 0 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
51024 2019-11-30 00:00:00 5.000000 6.000000 6.000000 3.000000 7.000000 4.000000 0.000000 18.000000 10.000000 13.000000 11.000000 157.000000 37.000000 0.000000 2.000000 1.000000 86.000000 14.000000 23.000000 161.000000 17.000000 10.000000 62.000000 6.000000 86.000000 89.000000 51.000000 105.000000 7.000000 13.000000 1.000000 87.000000 39.000000 46.000000 18.000000 58.000000 83.000000 14.000000 58.000000 4.000000 4.000000 115.000000 13.000000 7.000000 48.000000 101.000000 52.000000 101.000000 110.000000 18.000000 52.000000 0.000000 4.000000 83.000000 0.000000 9.000000 41.000000 2.000000 1.000000 3.000000 55.000000 187.000000 33.000000 10.000000 26.000000 81.000000 30.000000 46.000000 33.000000 44.000000 3.000000 27.000000 140.000000 1.000000 0.000000 1.000000 4.000000 10.000000 43.000000
51025 2019-11-30 00:30:00 11.000000 5.000000 5.000000 5.000000 3.000000 1.000000 0.000000 13.000000 6.000000 10.000000 6.000000 138.000000 18.000000 0.000000 2.000000 0.000000 66.000000 13.000000 20.000000 176.000000 9.000000 13.000000 66.000000 4.000000 52.000000 61.000000 38.000000 84.000000 2.000000 15.000000 0.000000 106.000000 22.000000 4.000000 20.000000 37.000000 67.000000 10.000000 66.000000 2.000000 5.000000 123.000000 5.000000 3.000000 65.000000 51.000000 32.000000 66.000000 104.000000 10.000000 47.000000 1.000000 3.000000 45.000000 7.000000 7.000000 30.000000 1.000000 4.000000 5.000000 36.000000 141.000000 38.000000 7.000000 22.000000 79.000000 22.000000 39.000000 27.000000 56.000000 3.000000 31.000000 124.000000 5.000000 0.000000 2.000000 7.000000 6.000000 48.000000
51026 2019-11-30 01:00:00 9.000000 7.000000 6.000000 6.000000 5.000000 1.000000 0.000000 12.000000 4.000000 10.000000 11.000000 118.000000 25.000000 0.000000 2.000000 0.000000 62.000000 7.000000 21.000000 168.000000 8.000000 5.000000 61.000000 1.000000 31.000000 48.000000 27.000000 89.000000 6.000000 23.000000 4.000000 50.000000 31.000000 0.000000 12.000000 37.000000 49.000000 3.000000 42.000000 3.000000 2.000000 141.000000 14.000000 3.000000 87.000000 35.000000 27.000000 45.000000 60.000000 8.000000 45.000000 4.000000 4.000000 76.000000 0.000000 11.000000 31.000000 0.000000 2.000000 3.000000 39.000000 107.000000 42.000000 7.000000 15.000000 56.000000 20.000000 36.000000 24.000000 27.000000 3.000000 31.000000 107.000000 5.000000 2.000000 1.000000 4.000000 5.000000 38.000000
51027 2019-11-30 01:30:00 10.000000 5.000000 4.000000 6.000000 1.000000 1.000000 0.000000 8.000000 3.000000 7.000000 9.000000 101.000000 13.000000 0.000000 2.000000 0.000000 79.000000 5.000000 9.000000 172.000000 6.000000 3.000000 56.000000 2.000000 44.000000 44.000000 29.000000 114.000000 3.000000 16.000000 5.000000 20.000000 20.000000 0.000000 7.000000 39.000000 29.000000 4.000000 59.000000 2.000000 2.000000 136.000000 7.000000 2.000000 89.000000 28.000000 31.000000 42.000000 66.000000 6.000000 25.000000 2.000000 2.000000 70.000000 4.000000 4.000000 33.000000 1.000000 3.000000 7.000000 31.000000 74.000000 37.000000 10.000000 15.000000 49.000000 21.000000 29.000000 21.000000 26.000000 1.000000 43.000000 95.000000 4.000000 7.000000 1.000000 4.000000 2.000000 31.000000
51028 2019-11-30 02:00:00 13.000000 9.000000 0.000000 4.000000 3.000000 1.000000 0.000000 6.000000 7.000000 7.000000 5.000000 83.000000 15.000000 0.000000 3.000000 0.000000 68.000000 5.000000 5.000000 131.000000 8.000000 3.000000 49.000000 0.000000 32.000000 44.000000 26.000000 103.000000 2.000000 15.000000 2.000000 21.000000 12.000000 0.000000 6.000000 35.000000 17.000000 4.000000 69.000000 3.000000 3.000000 161.000000 6.000000 0.000000 67.000000 14.000000 20.000000 34.000000 80.000000 5.000000 27.000000 2.000000 3.000000 44.000000 0.000000 7.000000 18.000000 1.000000 1.000000 9.000000 25.000000 83.000000 30.000000 9.000000 13.000000 39.000000 14.000000 9.000000 18.000000 9.000000 3.000000 67.000000 85.000000 7.000000 4.000000 3.000000 5.000000 3.000000 28.000000
weather[weather["ds"]<="2019-11-30 02:00:00"].tail(5)
ds REPORT_TYPE SOURCE HourlyAltimeterSetting HourlyDewPointTemperature HourlyDryBulbTemperature HourlyPrecipitation HourlyPresentWeatherType HourlyPressureChange HourlyPressureTendency HourlyRelativeHumidity HourlySkyConditions HourlySeaLevelPressure HourlyStationPressure HourlyVisibility HourlyWetBulbTemperature HourlyWindDirection HourlyWindGustSpeed HourlyWindSpeed REM
39951 2019-11-29 22:51:00 FM-15 7 30.11 23 37 0.00 NaN NaN NaN 57.000000 BKN:07 140 OVC:08 250 30.11 30.08 10.00 32.000000 340 NaN 11.000000 MET10111/29/19 22:51:01 METAR KLGA 300351Z 340...
39952 2019-11-29 23:51:00 FM-15 7 30.09 22 36 0.00 NaN NaN NaN 57.000000 OVC:08 150 30.09 30.06 10.00 31.000000 350 NaN 10.000000 MET10411/29/19 23:51:02 METAR KLGA 300451Z 350...
39953 2019-11-30 00:51:00 FM-15 7 30.08 21 36 0.00 NaN 0.040000 6.000000 55.000000 OVC:08 150 30.08 30.05 10.00 31.000000 360 NaN 11.000000 MET11211/30/19 00:51:02 METAR KLGA 300551Z 360...
39954 2019-11-30 01:00:00 FM-12 4 NaN 21 36 NaN NaN 0.040000 6.000000 55.000000 NaN 30.08 30.04 9.94 31.000000 360 NaN 11.000000 SYN08072503 32966 83610 10022 21061 30172 4018...
39955 2019-11-30 01:51:00 FM-15 7 30.08 20 36 0.00 NaN NaN NaN 52.000000 OVC:08 150 30.08 30.05 10.00 30.000000 350 NaN 15.000000 MET09911/30/19 01:51:02 METAR KLGA 300651Z 350...

上記データを例として、2019-11-30 01:00:00の予測をする時

  • 天気情報の39,953行目の「2019-11-30 00:51:00」までのデータを使用するのはOKでしょうか。
  • 天気情報の39,954行目の「2019-11-30 01:00:00」以降のデータを使用するのはNGでしょうか。
    ご回答よろしくお願いいたします。

〜〜質問はここまで〜〜

以下、直前のデータまで使用できると仮定した場合の集計と可視化サンプル置いておきます。
非常に雑な集計で申し訳ありません。ご参考までに。

# 縦持ち
data_df = train.copy()
data_df = pd.melt(
    data_df, id_vars=["ds"], value_vars=list(map(str, range(79))),
    var_name="id", value_name="y"
)
data_df.loc[:,"id"] = data_df["id"].astype("int64")

# お天気情報merge用の列追加
data_df.loc[:,"ds_h"] = pd.to_datetime(data_df["ds"].dt.strftime('%Y-%m-%d %H')+":00:00")
data_df
ds id y ds_h
0 2017-01-01 00:00:00 0 53.000000 2017-01-01 00:00:00
1 2017-01-01 00:30:00 0 83.000000 2017-01-01 00:00:00
2 2017-01-01 01:00:00 0 69.000000 2017-01-01 01:00:00
3 2017-01-01 01:30:00 0 76.000000 2017-01-01 01:00:00
4 2017-01-01 02:00:00 0 101.000000 2017-01-01 02:00:00
... ... ... ... ...
4034683 2019-11-30 21:30:00 78 75.000000 2019-11-30 21:00:00
4034684 2019-11-30 22:00:00 78 85.000000 2019-11-30 22:00:00
4034685 2019-11-30 22:30:00 78 72.000000 2019-11-30 22:00:00
4034686 2019-11-30 23:00:00 78 88.000000 2019-11-30 23:00:00
4034687 2019-11-30 23:30:00 78 78.000000 2019-11-30 23:00:00

4034688 rows × 4 columns

weather.loc[:,"year"] = weather["ds"].dt.year
weather.loc[:,"month"] = weather["ds"].dt.month
weather.loc[:,"day"] = weather["ds"].dt.day
weather.loc[:,"hour"] = weather["ds"].dt.hour
weather.loc[:,"minute"] = weather["ds"].dt.minute

weather.loc[:,"ds_h"] = pd.to_datetime(
    weather["year"].astype(str) +"-"+ weather["month"].astype(str).str.zfill(2) +"-"+ weather["day"].astype(str).str.zfill(2) +
    " "+ weather["hour"].astype(str).str.zfill(2) + ":00:00"
)
# 全部一時間後に参照できるようにする
weather.loc[:,"ds_h"] = weather["ds_h"] + datetime.timedelta(hours=+1)
weather[["ds","year","month","day","hour","minute","ds_h"]].head(10)
ds year month day hour minute ds_h
0 2017-01-01 00:51:00 2017 1 1 0 51 2017-01-01 01:00:00
1 2017-01-01 01:00:00 2017 1 1 1 0 2017-01-01 02:00:00
2 2017-01-01 01:51:00 2017 1 1 1 51 2017-01-01 02:00:00
3 2017-01-01 02:51:00 2017 1 1 2 51 2017-01-01 03:00:00
4 2017-01-01 03:51:00 2017 1 1 3 51 2017-01-01 04:00:00
5 2017-01-01 04:00:00 2017 1 1 4 0 2017-01-01 05:00:00
6 2017-01-01 04:51:00 2017 1 1 4 51 2017-01-01 05:00:00
7 2017-01-01 05:51:00 2017 1 1 5 51 2017-01-01 06:00:00
8 2017-01-01 06:51:00 2017 1 1 6 51 2017-01-01 07:00:00
9 2017-01-01 07:00:00 2017 1 1 7 0 2017-01-01 08:00:00
col_weather = [
  "HourlyDewPointTemperature","HourlyDryBulbTemperature","HourlyPrecipitation","HourlyPressureChange","HourlyPressureTendency","HourlyRelativeHumidity",
  "HourlySkyConditions","HourlySeaLevelPressure","HourlyStationPressure","HourlyVisibility","HourlyWetBulbTemperature","HourlyWindDirection","HourlyWindGustSpeed","HourlyWindSpeed",
]
# 強引にfloat変換(変換できないところはNA)
for c in col_weather:
  weather.loc[:,c] = pd.to_numeric(weather[c], errors="coerce")

# 時間ごとに平均
weather_ = pd.DataFrame({
  "ds_h":pd.date_range(start='2017-01-01 00:00:00', end='2019-12-07 23:00:00', freq='60T')
})
weather_ = pd.merge(
    weather_,
    weather[["ds_h"]+col_weather].groupby(["ds_h"]).mean().fillna(0).reset_index(),
    how="left", on="ds_h"
).set_index("ds_h")
# NAは線形補完
weather_ = weather_.interpolate(method='linear', axis=0)
weather_ = weather_.reset_index()
weather_.head(5)
# 一番最初のデータがなくなったけど、まぁいいか
ds_h HourlyDewPointTemperature HourlyDryBulbTemperature HourlyPrecipitation HourlyPressureChange HourlyPressureTendency HourlyRelativeHumidity HourlySkyConditions HourlySeaLevelPressure HourlyStationPressure HourlyVisibility HourlyWetBulbTemperature HourlyWindDirection HourlyWindGustSpeed HourlyWindSpeed
0 2017-01-01 00:00:00 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
1 2017-01-01 01:00:00 25.000000 46.000000 0.000000 0.020000 6.000000 44.000000 0.000000 29.890000 29.860000 10.000000 38.000000 230.000000 0.000000 10.000000
2 2017-01-01 02:00:00 26.500000 45.500000 0.000000 0.020000 6.000000 48.000000 41.000000 29.895000 29.860000 9.970000 38.000000 230.000000 0.000000 9.000000
3 2017-01-01 03:00:00 26.000000 46.000000 0.000000 0.000000 0.000000 46.000000 0.000000 29.910000 29.880000 10.000000 38.000000 240.000000 0.000000 13.000000
4 2017-01-01 04:00:00 25.000000 46.000000 0.000000 -0.040000 3.000000 44.000000 0.000000 29.930000 29.900000 10.000000 38.000000 240.000000 25.000000 10.000000
data_df_w = pd.merge(data_df, weather_, how="left", on="ds_h")
data_df_w.head(5)
ds id y ds_h HourlyDewPointTemperature HourlyDryBulbTemperature HourlyPrecipitation HourlyPressureChange HourlyPressureTendency HourlyRelativeHumidity HourlySkyConditions HourlySeaLevelPressure HourlyStationPressure HourlyVisibility HourlyWetBulbTemperature HourlyWindDirection HourlyWindGustSpeed HourlyWindSpeed
0 2017-01-01 00:00:00 0 53.000000 2017-01-01 00:00:00 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
1 2017-01-01 00:30:00 0 83.000000 2017-01-01 00:00:00 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
2 2017-01-01 01:00:00 0 69.000000 2017-01-01 01:00:00 25.000000 46.000000 0.000000 0.020000 6.000000 44.000000 0.000000 29.890000 29.860000 10.000000 38.000000 230.000000 0.000000 10.000000
3 2017-01-01 01:30:00 0 76.000000 2017-01-01 01:00:00 25.000000 46.000000 0.000000 0.020000 6.000000 44.000000 0.000000 29.890000 29.860000 10.000000 38.000000 230.000000 0.000000 10.000000
4 2017-01-01 02:00:00 0 101.000000 2017-01-01 02:00:00 26.500000 45.500000 0.000000 0.020000 6.000000 48.000000 41.000000 29.895000 29.860000 9.970000 38.000000 230.000000 0.000000 9.000000
# なんとなくこの期間を可視化
plot_d = data_df_w[(data_df_w["ds"]>="2018-11-08")&(data_df_w["ds"]<="2018-11-20")]

fig, ax = plt.subplots(len(col_weather), 1, figsize=(20, len(col_weather)*4))

i = 31 # なんとなく31地点を可視化(空港?)
for j, col in enumerate(col_weather):
  ax2 = ax[j].twinx()
  ax[j].plot(plot_d[plot_d["id"]==i]['ds'], plot_d[plot_d["id"]==i]["y"], color=cm.tab20.colors[0])
  ax2.plot(plot_d[plot_d["id"]==i]['ds'], plot_d[plot_d["id"]==i][col], color=cm.tab20.colors[j+1])
  ax[j].set_title(str(i)+"_"+col)

  # 色変更
  ax2.spines['left'].set_color(cm.tab20.colors[0])
  ax2.spines['right'].set_color(cm.tab20.colors[j+1])
  ax[j].tick_params(axis='y', colors=cm.tab20.colors[0])
  ax2.tick_params(axis='y', colors=cm.tab20.colors[j+1])

plt.show()
plt.close()
plt.clf()
Output hidden; open in https://colab.research.google.com to view.

添付データ

  • fig_weather_31.png?X-Amz-Expires=10800&X-Amz-Date=20241123T042622Z&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIP7GCBGMWPMZ42PQ
  • Aws4 request&x amz signedheaders=host&x amz signature=4cfdc25d48bed85701b4f2cec354c299d5227f2875d60e31387ea6af5b379bed
    ProbSpace_official

    mini_forest様

    ご質問いただき、ありがとうございます。ProbSpace運営事務局です。
    天気データに関してですが、予測の際に、未来情報をご使用されることは問題ございません。

    タクシーの利用判断におきましては、直近の天気予報情報が参考として用いられる可能性がございます。
    提供されているデータは実績値であり、予報情報とは異なりますが、近似的な意味での参考データとしてご使用いただけます。

    ご確認のほど、よろしくお願い申し上げます。

    Aws4 request&x amz signedheaders=host&x amz signature=c488fe242f155db232ef752a09f407b76fbc98629b3f90a8a21e7515e30f1616
    mini_forest

    運営担当者様

    ご回答ありがとうございます。
    実績値であるが、予報情報と捉えて未来の値も使用可能であること承知いたしました。

    以上、よろしくお願いいたします。

    Aws4 request&x amz signedheaders=host&x amz signature=4cfdc25d48bed85701b4f2cec354c299d5227f2875d60e31387ea6af5b379bed
    ProbSpace_official

    mini_forest様

    また何か疑問点などございましたらお問い合わせください。
    引き続き、よろしくお願いいたします。

    Favicon
    new user
    コメントするには 新規登録 もしくは ログイン が必要です。