{"id":4010,"date":"2023-11-10T18:01:55","date_gmt":"2023-11-10T09:01:55","guid":{"rendered":"https:\/\/blog.since2020.jp\/?p=4010"},"modified":"2024-02-27T18:25:40","modified_gmt":"2024-02-27T09:25:40","slug":"%e3%80%9015%e3%81%ae%e7%b2%be%e5%ba%a6%e5%90%91%e4%b8%8a%e3%80%91amzon%e3%81%8c%e9%96%8b%e7%99%ba%e3%81%97%e3%81%9f%e6%99%82%e7%b3%bb%e5%88%97%e3%83%a2%e3%83%87%e3%83%ab%e3%80%81deepar%e3%81%8c","status":"publish","type":"post","link":"https:\/\/since2020.jp\/media\/%e3%80%9015%e3%81%ae%e7%b2%be%e5%ba%a6%e5%90%91%e4%b8%8a%e3%80%91amzon%e3%81%8c%e9%96%8b%e7%99%ba%e3%81%97%e3%81%9f%e6%99%82%e7%b3%bb%e5%88%97%e3%83%a2%e3%83%87%e3%83%ab%e3%80%81deepar%e3%81%8c\/","title":{"rendered":"\u301015%\u306e\u7cbe\u5ea6\u5411\u4e0a\u3011Amazon\u304c\u958b\u767a\u3057\u305f\u6642\u7cfb\u5217\u30e2\u30c7\u30eb\u3001DeepAR\u304c\u51c4\u3044"},"content":{"rendered":"\n

Amazon\u304c\u958b\u767a\u3057\u305f\u6642\u7cfb\u5217\u30e2\u30c7\u30ebDeepAR\u306b\u3064\u3044\u3066\u89e3\u8aac\u3057\u3066\u3044\u307e\u3059\u3002\r\n\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3084\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc\u306e\u8a2d\u5b9a\u3001\u7279\u5fb4\u3084\u30e1\u30ea\u30c3\u30c8\u306b\u3064\u3044\u3066\u8a73\u3057\u304f\u77e5\u308c\u308b\u8a18\u4e8b\u306b\u306a\u3063\u3066\u3044\u307e\u3059\u3002\r\n<\/p>\n\n\n

DeepAR\u306f2019\u5e74\u306bAmazon\u304c\u958b\u767a\u3057\u305f\u6642\u7cfb\u5217\u30e2\u30c7\u30eb<\/h2>\n

Amazon\u304c\u958b\u767a\u3057\u305fDeepAR\u306f\u3001\u6df1\u5c64\u5b66\u7fd2\u3092\u4f7f\u7528\u3057\u3066\u6642\u7cfb\u5217\u3092\u4e88\u6e2c\u3059\u308b\u305f\u3081\u306e\u6559\u5e2b\u4ed8\u304d\u5b66\u7fd2\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0<\/span>\u3067\u3059\u3002\u8ad6\u6587\u306e\u6295\u7a3f\u5e74\u306f2019\u5e74\u306a\u306e\u3067\u6bd4\u8f03\u7684\u6700\u8fd1\u306e\u30e2\u30c7\u30eb\u3067\u3059\u306d\u3002\"\"<\/p>\r\n

\u5f15\u7528\uff1aSalinas, D., Flunkert, V., & Gasthaus, J. (2019). DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks. arXiv:1704.04110v3<\/em>. Amazon Research Germany.<\/span><\/p>\r\n

DeepAR\u306f\u3001\u8907\u6570\u306e\u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u3092\u8003\u616e\u3057\u306a\u304c\u3089\u3001\u9ad8\u3044\u4e88\u6e2c\u7cbe\u5ea6\u3092\u76ee\u6307\u3059\u3082\u306e\u3068\u3057\u3066\u8a2d\u8a08\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u4ee5\u4e0b\u306b\u3001DeepAR\u306e\u4e3b\u8981\u306a\u7279\u5fb4\u3068\u30e1\u30ea\u30c3\u30c8\u306b\u3064\u3044\u3066\u89e3\u8aac\u3057\u307e\u3059\u3002<\/p>\r\n

DeepAR\u306f\u3001Amazon\u306e\u30af\u30e9\u30a6\u30c9\u6a5f\u68b0\u5b66\u7fd2\u30b5\u30fc\u30d3\u30b9\u3067\u3042\u308bAmazon SageMaker\u3067\u5229\u7528\u53ef\u80fd\u3067\u3059\u3002<\/p>\r\n

\u79c1\u306fDataiku\u3067\u4f7f\u7528\u3057\u3066\u521d\u3081\u3066DeepAR\u306e\u5b58\u5728\u3092\u77e5\u308a\u307e\u3057\u305f\u3002<\/p>\r\n

<\/p>\n\n

DeepAR\u306e\u30e1\u30ea\u30c3\u30c8\uff08\u9b45\u529b\u70b9\uff09<\/h2>\n

\uff11\uff0e\u5916\u90e8\u306e\u5909\u6570\u3092\u8003\u616e\u53ef\u80fd<\/strong><\/p>\r\n

\u65e5\u4ed8\u30c7\u30fc\u30bf\u3068\u76ee\u7684\u5909\u6570\u306e\u307f\u3067\u306f\u306a\u304f\u3001\u4ed6\u306b\u95a2\u9023\u3057\u305f\u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u3082\u7279\u5fb4\u91cf\u3068\u3057\u3066\u7d44\u307f\u8fbc\u307f\u53ef\u80fd\u3067\u3059\u3002<\/p>\r\n

\uff12\uff0e\u5916\u90e8\u5909\u6570\u30c7\u30fc\u30bf\u304c\u4f55\u767e\u3082\u542b\u307e\u308c\u3066\u3044\u308b\u5834\u5408\u306b\u771f\u4fa1\u3092\u767a\u63ee\u3059\u308b<\/strong><\/p>\r\n

\uff5c\u53ef\u80fd\u306a\u9650\u308a\u591a\u304f\u306e\u6642\u7cfb\u5217\u306b\u3064\u3044\u3066 DeepAR \u30e2\u30c7\u30eb\u3092\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u3053\u3068\u3092\u304a\u52e7\u3081\u3057\u307e\u3059\u3002<\/p>\r\n

\uff5c\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306b\u95a2\u9023\u3059\u308b\u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u304c\u4f55\u767e\u3082\u542b\u307e\u308c\u3066\u3044\u308b\u5834\u5408\u3001DeepAR \u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306f\u6a19\u6e96\u306e\u65b9\u6cd5\u3088\u308a\u512a\u308c\u305f\u6027\u80fd\u3092\u767a\u63ee\u3059\u308b\u3088\u3046\u306b\u306a\u308a\u307e\u3059\u3002<\/p>\r\n

\uff5c<\/strong>\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306b\u4f55\u767e\u3082\u306e\u95a2\u9023\u3059\u308b\u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u304c\u542b\u307e\u308c\u3066\u3044\u308b\u5834\u5408\u3001DeepAR \u306f\u6a19\u6e96\u306e ARIMA \u3084 ETS \u30e1\u30bd\u30c3\u30c9\u3088\u308a\u3082\u512a\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\r\n

\u5f15\u7528\uff1aDeepAR \u4e88\u6e2c\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0 – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/span><\/p>\r\n

ARIMA\u30e2\u30c7\u30eb\u3088\u308a\u512a\u308c\u3066\u3044\u308b\u5ba3\u8a00\u3057\u3066\u3044\u307e\u3059\u306d\u7b11<\/p>\r\n

\uff13\uff0e\u6700\u65b0\u306e\u65b9\u6cd5\u3068\u6bd4\u8f03\u3057\u3066\u7cbe\u5ea6\u304c\u7d0415%\u5411\u4e0a<\/strong><\/p>\r\n

\uff5cWe show through extensive empirical evaluation on several real-world forecasting data sets accuracy improvements of around 15% compared to state-of-the-art methods.<\/p>\r\n

\u5f15\u7528\uff1aSalinas, D., Flunkert, V., & Gasthaus, J. (2019). DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks. arXiv:1704.04110v3<\/em>. Amazon Research Germany.<\/span><\/p>\r\n

\u3053\u308c\u306f\u9b45\u529b\u7684\u3067\u3059\u3088\u306d\u3002<\/p>\r\n

2019\u5e74\u6642\u70b9\u3067\u306e\u8a71\u3067\u306f\u3042\u308a\u307e\u3059\u304c\u3001<\/p>\r\n

\uff14\uff0e\u78ba\u7387\u7684\u4e88\u6e2c\u304c\u53ef\u80fd<\/strong><\/p>\r\n

DeepAR\u306f\u3001\u4e88\u6e2c\u306e\u4e0d\u78ba\u5b9f\u6027\u3092\u8003\u616e\u3057\u305f\u78ba\u7387\u7684\u306a\u4e88\u6e2c\u3092\u63d0\u4f9b\u3057\u307e\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u3001\u4e88\u6e2c\u306e\u4e2d\u592e\u5024\u3060\u3051\u3067\u306a\u304f\u3001\u4fe1\u983c\u533a\u9593\u3084\u5206\u5e03\u3092\u5f97\u308b\u3053\u3068\u304c\u3067\u304d\u3001\u30ea\u30b9\u30af\u306e\u8a55\u4fa1\u3084\u610f\u601d\u6c7a\u5b9a\u306b\u6709\u76ca\u306a\u60c5\u5831\u3092\u63d0\u4f9b\u3057\u307e\u3059\u3002<\/p>\r\n

<\/p>\n\n

DeepAR\u306e\u30c7\u30e1\u30ea\u30c3\u30c8<\/h2>\n

\uff11\uff0e\u6df1\u5c64\u5b66\u7fd2\u3092\u4f7f\u7528\u3057\u3066\u3044\u308b\u306e\u3067\u3001\u8aac\u660e\u900f\u660e\u6027\u304c\u306a\u3044<\/strong><\/p>\r\n

\u3053\u3053\u3067\u3059\u306d\uff5e\u3002\u3057\u3087\u3046\u304c\u306a\u3044\u3093\u3067\u3059\u3051\u3069\u306d\u3002<\/p>\r\n

\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u3084LightGBM\u306a\u3069\u3068\u9055\u3063\u3066\u3001\u91cd\u8981\u306a\u7279\u5fb4\u91cf\u306a\u3069\u8aac\u660e\u304c\u96e3\u3057\u3044\u3067\u3059\u3002<\/p>\r\n

\u307e\u305f\u3001\u3069\u3046\u3057\u3066\u3053\u306e\u3088\u3046\u306a\u7d50\u679c\u306b\u306a\u3063\u305f\u306e\u304b\u8aac\u660e\u3059\u308b\u3053\u3068\u304c\u96e3\u3057\u3044\u3067\u3059\u3002<\/p>\r\n

\uff12\uff0e\u8aac\u660e\u5909\u6570\u304c\u5c11\u306a\u3044\u5834\u5408\u3001\u771f\u4fa1\u3092\u767a\u63ee\u3057\u306a\u3044\u53ef\u80fd\u6027\u304c\u3042\u308b<\/strong><\/p>\r\n

\u30e1\u30ea\u30c3\u30c8\u306e\u9006\u3067\u3001\u5341\u5206\u306a\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u304c\u306a\u3044\u5834\u5408\u7cbe\u5ea6\u3092\u767a\u63ee\u3057\u306a\u3044\u304b\u3082\u3057\u308c\u307e\u305b\u3093\u3002<\/p>\r\n

<\/p>\n\n

DeepAR\u306e\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0<\/h2>\n

\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u304b\u3089\u30e9\u30f3\u30c0\u30e0\u306b\u30b5\u30f3\u30d7\u30ea\u30f3\u30b0\u3059\u308b\u3053\u3068\u3067\u5b66\u7fd2\u3059\u308b<\/strong><\/p>\r\n

\"\"<\/p>\r\n

\uff5cDeepAR \u306f\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u5404\u6642\u7cfb\u5217\u304b\u3089\u3044\u304f\u3064\u304b\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u4f8b\u3092\u30e9\u30f3\u30c0\u30e0\u306b\u30b5\u30f3\u30d7\u30ea\u30f3\u30b0\u3059\u308b\u3053\u3068\u306b\u3088\u3063\u3066\u30e2\u30c7\u30eb\u3092\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3057\u307e\u3059\u3002\u5404\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u4f8b\u306f\u3001\u4e8b\u524d\u5b9a\u7fa9\u3055\u308c\u305f\u56fa\u5b9a\u9577\u3092\u6301\u3064\u4e00\u5bfe\u306e\u96a3\u63a5\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u3068\u4e88\u6e2c\u30a6\u30a3\u30f3\u30c9\u30a6\u3067\u69cb\u6210\u3055\u308c\u3066\u3044\u307e\u3059\u3002context_length<\/span>\u00a0\u30cf\u30a4\u30d1\u30fc\u30d1\u30e9\u30e1\u30fc\u30bf\u306f\u3001\u3069\u306e\u7a0b\u5ea6\u306e\u904e\u53bb\u307e\u3067\u9061\u3063\u3066\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u3092\u8abf\u3079\u3089\u308c\u308b\u306e\u304b\u3092\u5236\u5fa1\u3057\u3001prediction_length<\/span>\u00a0\u30cf\u30a4\u30d1\u30fc\u30d1\u30e9\u30e1\u30fc\u30bf\u306f\u3001\u3069\u306e\u7a0b\u5ea6\u306e\u672a\u6765\u307e\u3067\u4e88\u6e2c\u3092\u751f\u6210\u3067\u304d\u308b\u304b\u3092\u5236\u5fa1\u3057\u307e\u3059\u3002<\/p>\r\n

\u5f15\u7528\uff1aDeepAR \u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306e\u4ed5\u7d44\u307f – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/span><\/p>\r\n

\u3053\u306e\u56f3\u3067\u3044\u3046\u3068\u3001\u7dd1\u306e\u671f\u9593\u306e\u90e8\u5206\u3092\u30e9\u30f3\u30c0\u30e0\u306b\u5b9a\u3081\u3001\u9752\u306b\u8a72\u5f53\u3059\u308b\u4e88\u6e2c\u671f\u9593\u3067\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u3089\u3057\u3044\u3067\u3059\u3002<\/p>\r\n

\u3053\u3053\u3067context_length<\/code>\u00a0\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc\u306f\u3069\u306e\u7a0b\u5ea6\u904e\u53bb\u306e\u671f\u9593\u3092\u30b5\u30f3\u30d7\u30ea\u30f3\u30b0\u3057\u3001\u5b66\u7fd2\u3055\u305b\u308b\u304b\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc\u3068\u3044\u3046\u3053\u3068\u306b\u306a\u308a\u307e\u3059\u3002<\/p>\r\n

\u307e\u305f\u3001prediction_length<\/code>\u00a0\u3067\u306f\u3001\u4e88\u6e2c\u671f\u9593\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc\u3067\u3059\u306d\u3002<\/p>\r\n

\u5b63\u7bc0\u6027\u3084\u30c8\u30ec\u30f3\u30c9\u3092\u8003\u616e\u3059\u308b\u305f\u3081\u3001\u30bf\u30fc\u30b2\u30c3\u30c8\u6642\u7cfb\u5217\u304b\u3089\u904e\u53bb\u306e\u9045\u5ef6\u3057\u305f\u5024\u3082\u8003\u616e\u3059\u308b<\/strong><\/p>\r\n

\"\"\uff5c\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3055\u308c\u305f\u30e2\u30c7\u30eb\u306f\u3001\u63a8\u8ad6\u306e\u305f\u3081\u306b\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u4e2d\u306b\u4f7f\u7528\u3055\u308c\u305f\u304b\u3069\u3046\u304b\u306b\u304b\u304b\u308f\u3089\u305a\u3001\u30bf\u30fc\u30b2\u30c3\u30c8\u6642\u7cfb\u5217\u3092\u5165\u529b\u3068\u3057\u3066\u53d7\u3051\u53d6\u308a\u3001\u305d\u308c\u4ee5\u964d\u306e\u00a0prediction_length<\/code>\u00a0\u5024\u306e\u78ba\u7387\u5206\u5e03\u3092\u4e88\u6e2c\u3057\u307e\u3059\u3002<\/p>\r\n

\u5f15\u7528\uff1aDeepAR \u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306e\u4ed5\u7d44\u307f – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/span><\/p>\r\n

\u4e0a\u8a18\u306e\u56f3\u306e\u4f8b\u3067\u3059\u3068\u3001t\u6642\u70b9\u306e\u5024\u306e\u307f\u3067\u306f\u306a\u304f\u3001\u5b63\u7bc0\u5468\u671f\u30921\u3068\u3057\u3001t-1,t-2,t-3\u6642\u70b9\u306a\u3069\u306e\u5024\u306e\u30d1\u30bf\u30fc\u30f3\u3082\u8003\u616e\u3059\u308b\u3089\u3057\u3044\u3067\u3059\u3002<\/p>\r\n

\u65b0\u305f\u306a\u6642\u7cfb\u5217\u7279\u5fb4\u91cf\u306e\u81ea\u52d5\u751f\u6210<\/strong><\/p>\r\n

\"\"<\/p>\r\n

\u5f15\u7528\uff1aDeepAR \u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306e\u4ed5\u7d44\u307f – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/span><\/p>\r\n

DeepAR\u3067\u306f\u4e00\u3064\u306e\u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u304b\u3089\u7279\u5fb4\u3092\u81ea\u52d5\u7684\u306b\u62bd\u51fa\u3057\u3001\u65b0\u305f\u306a\u7279\u5fb4\u91cf\u3092\u4f5c\u6210\u3057\u3066\u304f\u308c\u307e\u3059\u3002<\/p>\r\n

\u4e0a\u306e\u56f3\u3067\u306f\u3001\u5143\u30c7\u30fc\u30bf\u304b\u3089\u6d3e\u751f\u3057\u305f 2 \u3064\u306e\u6642\u7cfb\u5217\u306e\u7279\u5fb4\u3092\u793a\u3057\u3066\u3044\u307e\u3059\u3002ui,1,t<\/em>\u00a0\u306f\u305d\u306e\u65e5\u306e\u6642\u9593\u3092\u8868\u3057\u3001ui,2,t<\/em>\u00a0\u306f\u305d\u306e\u66dc\u65e5\u3092\u8868\u3057\u307e\u3059\u3002<\/p>\r\n

\u4f8b\u3048\u3070\u3001\u6642\u7cfb\u5217\u306e\u7c92\u5ea6\u304c\u5206\u5358\u4f4d\u3060\u3063\u305f\u5834\u5408\u306b\u4ee5\u4e0b\u306e\u7279\u5fb4\u91cf\u304c\u751f\u6210\u3055\u308c\u307e\u3059\u3002<\/p>\r\n\r\n\r\n\r\n\r\n\r\n
\u6642\u7cfb\u5217\u306e\u7c92\u5ea6<\/th>\r\n\u6d3e\u751f\u3059\u308b\u7279\u5fb4<\/th>\r\n<\/tr>\r\n<\/thead>\r\n
Minute<\/td>\r\nminute-of-hour,\u00a0hour-of-day,\u00a0day-of-week,\u00a0day-of-month,\u00a0day-of-year<\/td>\r\n<\/tr>\r\n<\/tbody>\r\n<\/table>\r\n

<\/p>\r\n

\u6df1\u5c64\u5b66\u7fd2\u30d9\u30fc\u30b9\u306e\u5b66\u7fd2<\/strong><\/p>\r\n

\"\"\uff5cFigure 2: Summary of the model. Training (left): At each time step t, the inputs to the network are the covariates xi,t, the target value at the previous time step zi,t\u22121, as well as the previous network output hi,t\u22121. The network output hi,t = h(hi,t\u22121, zi,t\u22121, xi,t, \u0398) is then used to compute the parameters \u03b8i,t = \u03b8(hi,t, \u0398) of the likelihood (z|\u03b8), which is used for training the model parameters. For prediction, the history of the time series zi,t is fed in for t < t0, then in the prediction range (right) for t \u2265 t0 a sample z\u02c6i,t \u223c<\/code> (\u00b7|\u03b8i,t) is drawn and fed back for the next point until the end of the prediction range t = t0 + T generating one sample trace. Repeating this prediction process yields many traces representing the joint predicted distribution.<\/p>\r\n

\u3053\u306e\u30e2\u30c7\u30eb\u306e\u6982\u8981\u306b\u57fa\u3065\u304f\u3068\u3001\u5404\u6642\u70b9t\u3067\u306e\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u3078\u306e\u5165\u529b\u306f\u4ee5\u4e0b\u306e\u30c7\u30fc\u30bf\u3067\u3059\uff1a<\/p>\r\n

1.\u5171\u5909\u91cf xi,t<\/strong><\/p>\r\n

\u3053\u308c\u306f\u5916\u90e8\u304b\u3089\u306e\u8ffd\u52a0\u60c5\u5831\u3084\u7279\u5fb4\u91cf\u3092\u6307\u3057\u307e\u3059\u3002\u4f8b\u3048\u3070\u3001\u6c17\u6e29\u3084\u30d7\u30ed\u30e2\u30fc\u30b7\u30e7\u30f3\u6d3b\u52d5\u306a\u3069\u3001\u76ee\u6a19\u5909\u6570\u306b\u5f71\u97ff\u3092\u4e0e\u3048\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u5916\u90e8\u306e\u5909\u6570\u3067\u3059\u3002<\/p>\r\n

2.\u524d\u306e\u6642\u70b9 t-1 \u306e\u76ee\u6a19\u5024 zi,t\u22121<\/strong><\/p>\r\n

\u3053\u308c\u306f\u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u306e\u524d\u306e\u5024\u3092\u6307\u3057\u3001\u30e2\u30c7\u30eb\u304c\u6b21\u306e\u6642\u70b9\u306e\u4e88\u6e2c\u3092\u884c\u3046\u969b\u306e\u53c2\u7167\u70b9\u3068\u3057\u3066\u4f7f\u7528\u3055\u308c\u307e\u3059\u3002<\/p>\r\n

3.\u524d\u306e\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u51fa\u529b hi,t\u22121<\/strong><\/p>\r\n

\u3053\u308c\u306f\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u304c\u524d\u306e\u6642\u70b9\u3067\u51fa\u529b\u3057\u305f\u5024\u3067\u3001\u6b21\u306e\u6642\u70b9\u306e\u4e88\u6e2c\u306b\u304a\u3044\u3066\u3082\u53c2\u8003\u3068\u3057\u3066\u7528\u3044\u3089\u308c\u307e\u3059\u3002<\/p>\r\n

\u3053\u308c\u3089\u306e\u5165\u529b\u3092\u57fa\u306b\u3001\u30e2\u30c7\u30eb\u306f\u6b21\u306e\u6642\u70b9t\u3067\u306e\u76ee\u6a19\u5909\u6570\u306e\u4e88\u6e2c\u3092\u884c\u3044\u307e\u3059<\/del>\u3002\uff08\u6b63\u78ba\u306b\u306f\u5206\u5e03\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc\u63a8\u5b9a\u3092\u884c\u3046\uff09\u305d\u3057\u3066\u3001\u4e88\u6e2c\u7bc4\u56f2\u5185\u3067\u306e\u3055\u3089\u306a\u308b\u4e88\u6e2c\u3092\u884c\u3046\u969b\u306b\u3001\u524d\u306e\u4e88\u6e2c\u5024\u304c\u65b0\u305f\u306a\u5165\u529b\u3068\u3057\u3066\u4f7f\u7528\u3055\u308c\u3001\u3053\u306e\u30d7\u30ed\u30bb\u30b9\u304c\u4e88\u6e2c\u7bc4\u56f2\u306e\u7d42\u308f\u308a\u307e\u3067\u7e70\u308a\u8fd4\u3055\u308c\u307e\u3059\u3002<\/p>\r\n

<\/p>\r\n

<\/p>\r\n

<\/p>\r\n

<\/p>\r\n

<\/p>\n\n

DeepAR\u306e\u30cf\u30a4\u30d1\u30fc\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc<\/h2>\n

\u4e3b\u8981\u306a\u3082\u306e\u3092\u7d39\u4ecb\u3057\u307e\u3059\u3002<\/p>\r\n

context_length\u30fb\u30fb\u30fb\u3069\u306e\u7a0b\u5ea6\u904e\u53bb\u306e\u671f\u9593\u3092\u9061\u3063\u3066\u30b5\u30f3\u30d7\u30ea\u30f3\u30b0\u3057\u3001\u5b66\u7fd2\u3055\u305b\u308b\u304b<\/p>\r\n

\u63a8\u5968\u7bc4\u56f2\uff1aMinValue: 1\u3001MaxValue: 200<\/p>\r\n

prediction_length\u30fb\u30fb\u30fb\u4e88\u6e2c\u671f\u9593<\/p>\r\n

epochs\u30fb\u30fb\u30fb\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u3078\u306e\u30d1\u30b9\u306e\u6700\u5927\u6570<\/p>\r\n

\u63a8\u5968\u7bc4\u56f2\uff1aMinValue: 1\u3001MaxValue: 1000<\/p>\r\n

time_freq\u30fb\u30fb\u30fb\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u5185\u306e\u6642\u7cfb\u5217\u306e\u8a73\u7d30\u5ea6<\/p>\r\n

cardinality\u30fb\u30fb\u30fb\u30ab\u30c6\u30b4\u30ea\u5225\u7279\u5fb4 (cat<\/code>) \u3092\u4f7f\u7528\u3059\u308b\u5834\u5408\u3001cardinality<\/code>\u00a0\u306f\u3001\u30ab\u30c6\u30b4\u30ea\u5225\u7279\u5fb4\u3054\u3068\u306e\u30ab\u30c6\u30b4\u30ea (\u30b0\u30eb\u30fc\u30d7) \u306e\u6570\u3092\u6307\u5b9a\u3059\u308b<\/p>\r\n

num_cells\u30fb\u30fb\u30fbRNN \u306e\u975e\u8868\u793a\u306e\u30ec\u30a4\u30e4\u30fc\u3054\u3068\u306b\u4f7f\u7528\u3059\u308b\u30bb\u30eb\u306e\u6570<\/p>\r\n

\u63a8\u5968\u7bc4\u56f2\uff1aMinValue: 30MaxValue: 200<\/p>\r\n

num_layers\u30fb\u30fb\u30fbRNN \u306e\u975e\u8868\u793a\u30ec\u30a4\u30e4\u30fc\u306e\u6570<\/p>\r\n

\u63a8\u5968\u7bc4\u56f2\uff1aMinValue: 1\u3001MaxValue: 8<\/p>\r\n

\u53c2\u8003\uff1aDeepAR \u30cf\u30a4\u30d1\u30fc\u30d1\u30e9\u30e1\u30fc\u30bf – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/span><\/p>\r\n

DeepAR \u30e2\u30c7\u30eb\u3092\u8abf\u6574\u3059\u308b – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/span><\/p>\r\n

<\/p>\n\n

\u53c2\u8003\u6587\u732e<\/h2>\n

DeepAR \u4e88\u6e2c\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0 – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/p>\r\n

DeepAR \u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306e\u4ed5\u7d44\u307f – \u30a2\u30de\u30be\u30f3 SageMaker (amazon.com)<\/a><\/p>\r\n

[1704.04110] DeepAR:\u81ea\u5df1\u56de\u5e30\u30ea\u30ab\u30ec\u30f3\u30c8\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306b\u3088\u308b\u78ba\u7387\u7684\u4e88\u6e2c (arxiv.org)<\/a><\/p>\r\n

<\/p>","protected":false},"excerpt":{"rendered":"

Amazon\u304c\u958b\u767a\u3057\u305f\u6642\u7cfb\u5217\u30e2\u30c7\u30ebDeepAR\u306b\u3064\u3044\u3066\u89e3\u8aac\u3057\u3066\u3044\u307e\u3059\u3002 \u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3084\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc\u306e\u8a2d\u5b9a\u3001\u7279\u5fb4\u3084\u30e1\u30ea\u30c3\u30c8\u306b\u3064\u3044\u3066\u8a73\u3057\u304f\u77e5\u308c\u308b\u8a18\u4e8b\u306b\u306a\u3063\u3066\u3044\u307e\u3059\u3002 DeepAR\u306f2019\u5e74\u306bAmazon\u304c\u958b\u767a\u3057\u305f\u6642\u7cfb\u5217\u30e2\u30c7 […]<\/p>\n","protected":false},"author":83,"featured_media":2525,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","swell_btn_cv_data":"","footnotes":"","_wp_rev_ctl_limit":""},"categories":[1246],"tags":[509,201,507,508],"class_list":["post-4010","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-data-infrastructure","tag-amazon-sagemaker","tag-aws","tag-deepar","tag-508"],"_links":{"self":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts\/4010","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/users\/83"}],"replies":[{"embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/comments?post=4010"}],"version-history":[{"count":1,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts\/4010\/revisions"}],"predecessor-version":[{"id":4409,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts\/4010\/revisions\/4409"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/media\/2525"}],"wp:attachment":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/media?parent=4010"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/categories?post=4010"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/tags?post=4010"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}