This research aimed to measure the effectiveness of Thai news headlines classification using an artificial neural network (ANN). The headlines consisted of i) political news, ii) sports news, iii) economic news, and iv) crime news, 1,200 headlines in total. The distribution of headlines was measured by using chi-square, information gain, and term frequency inverse class frequency (TFICF). Threshold default value was set in relation to terms of headlines before cross-validation was employed to categorize the data to examine the efficiency of the model using a neural network algorithm in classifying the headlines. The investigation of the news headline classification efficiency revealed that the 15-fold data division using TFICF was the most accurate in classifying headlines, with the accuracy rate of 99.60% and F-measure rate of 99.05%. Moreover, it was found that when more news headlines were provided as the learning data, the news headline classification became more accurate. Likewise, appropriate threshold value determination facilitated the selection of appropriate features in the headlines and resulted in more effective and accurate classification. Hence, it can be concluded that headline classification will be more accurate if the appropriate amount of learning data exists, and appropriate threshold value was set.

<span>This research aimed to create suitable forecasting models with long-short term memory (LSTM) from time series data, the price of rubber smoked sheets (RSS3) using 2,631 data from the Rubber Authority of Thailand for the past 10 years. The data was divided into two sets: first series 2,105 data points were used to create the LSTM prediction model; second series 526 data points were used to estimate forecasting performance using the root mean square error (RMSE), the mean absolute percentage error (MAPE), and accuracy rate of the model. The results showed that the most suitable forecasting model for time series data, with a total of 9 LSTM layers comprised of 3 primary LSTMs. Each LSTM layer has the number of neurons 100, 150, and 200 to obtain an optimal neural network of the LSTM technique. The number of epochs and iteration was 30, 40, and 50. Dropout layers between each LSTM layer have a probability of 30%. The results of the test to measure the performance of the time series forecasting data showed that the 9-layer model with the LSTM model architecture of LSTM 3 layers gave the best forecast, with RMSE of 2.4121, MAPE of 0.0413 and 95.88% accuracy rate.</span>