應用深度學習於Facebook廣告投放成效之預測

研究目的：本研究以Facebook廣告投放與深度學習的應用為研究議題，參考業者的行銷策略，進行深度學習模型之建構，並透過實驗設計，找出最佳模型參數值建立廣告成效預測模型。

研究方法：欲建立廣告成效預測模型，使用因素分析與迴歸分析，篩選出有顯著水準的輸入變數，並將輸入變數納入深度學習，並藉由實驗設計選出最佳模型參數值。

研究結果：本研究實驗分析結果得悉，經過三種實驗設計，選擇實驗一作為最佳模型參數值，並建立出廣告成效預測模型。

研究貢獻：本研究提出的廣告成效預測模型可根據目標受眾的設定預測受眾的點擊率，並根據點擊率的高低協助企業評估是否投放該支廣告，或者調整廣告投放參數以提高Facebook廣告投放之成效。

Purpose : This study investigated Facebook advertising through deep learning. In particular, it evaluates the advertising performance levels of the industry’s marketing strategies. To do so, a deep learning model was constructed, and an experiment was used to determine the model parameters that yield the best predictive performance.

Methods : To establish the deep learning model, factor analysis and regression analysis were used to determine the input variables of statistical significance to be incorporated into the deep learning model.

Result : Three experiments were conducted, and the first experiment was found to yield the best-performing model parameters, which were then used to establish the model for predicting advertising effectiveness.

Contribution : Specifically, this prediction model can predict the audience’s click-through rate based on the target audience’s settings. Companies can use this model for more effective Facebook advertising. Based on the click-through rate, companies can adjust serving parameters or determine whether Facebook advertising needs to be used.