以關鍵詞選擇用谷歌趨勢及ARIMAX於預測銷售

Keyword Selection for Google Trends in Forecasting Sales by ARIMAX

黃柏雄(Pao-Hsiung Huang)

指導教授 ： 白炳豐

國立暨南國際大學/管理學院/資訊管理學系/碩士(2016年)

摘要

什麼是專家？ 2007年，哈佛商業評論上發表文章“後天天才製造法”， 專家是“台上十分鐘，台下十年功。“（Ericsson, et al. 2007）。一位分析師可能是在預測的專家，但會遇到許多不同種類的產品和服務。很多時候要做出準確的評估將需要知道產品或服務。然而，需要大量的時間來調查的產品或服務。因此，我們可以通過選擇最相關的關鍵字的產品縮短時間並有準確的預測。我們的方法：（1）關於產品或服務，使用這些數據來提取關鍵字文本挖掘的評論; （2）使用谷歌趨勢進行預測的關鍵字; （3）隨著產品的銷售數據和谷歌趨勢數據，以獲得準確的銷售預測模型。我們使用評論來查找關鍵字的實驗結果比僅僅使用谷歌趨勢的數據的預測更準確。

關鍵字

預測銷售 ； 關鍵詞 ； 谷歌趨勢 ； Word2vec ； SARIMA ； ARIMAX

並列摘要

What is an expert? In 2007 Harvard Business Reviews published an article “The making of an expert,” “It takes time to become expert. Even the most gifted performers need a minimum of ten years of intense training before they win international competitions.” (Ericsson, et al. 2007). A forecast analyst could be experts at using the best modeling to make accurate predictions, but will face many different kinds of products, and services. Many times to make accurate assessment would require to know the product or services really well. However, it would require a lot of time to survey the product or service. Hence we come up with a method that could shorten the time for accurate forecasting by choosing the most relevant keywords for the product. Our task is performed in three steps: (1) Text mining reviews and/or blogs about the products or service, use the data to extract keywords; (2) Use the keywords in Google Trends for forecasting; (3) With the product’s sales data and Google Trends data to get an accurate sale forecast model. Our experimental results using text mining data to find keywords is more accurate than just using Google’s own data for forecasting.

並列關鍵字

Forecast of sales ； Keyword Selection ； Google Trends ； Word2vec ； SARIMA ； ARIMAX

參考文獻

Agrawal, D., Schorling, C., 1996. Market share forecasting: an empirical comparison of artificial neural networks and multinomial logit model. J. Retail. 72 (4), 383–407.

Ahluwalia, R. (2002). How prevalent is the negativity effect in consumer environments?. Journal of consumer research, 29(2), 270-279.

Alexandridis, A.K., Zapranis, A.D., 2013. Wavelet neural networks: a practical guide. Neural Netw. 42, 1–27.

Alon, I., Qi, M., Sadowski, R.J., 2001. Forecasting aggregate retail sales: a comparison of artificial neural networks and traditional methods. J. Retail. Consum. Serv. 8(3), 147–156.

Amazon. (2016). Amazon iPhone lists. Retrieved from https://www.amazon.com/s/ref=sr_nr_n_0?fst=as%3Aoff&rh=n%3A2335752011%2Cn%3A7072561011%2Ck%3AiPhone%2Cp_89%3AApple&keywords=iPhone&ie=UTF8&qid=1471494002&rnid=2335753011. July 2016