Hajj Dataset 2008-2019: Ministry of Religious Affairs Malang City
Description
This dataset contains administrative information of Hajj registrants recorded by the Ministry of Religious Affairs in Malang City from 2008 to 2019. It includes various important attributes such as age at registration, registration year, waiting time estimation, departure quotas, and predicted departure years based on two approaches: conventional estimation and Long Short-Term Memory (LSTM)-based prediction. The dataset has undergone preprocessing steps such as missing value handling, data type analysis, normalization, and manual data splitting for training and testing. It is specifically prepared to support predictive modeling tasks in the context of religious public services. Researchers can use this dataset for studies related to time series prediction, social policy planning, machine learning applications in government systems, and decision support systems for religious affairs.
Files
Steps to reproduce
1. Data Collection: > Raw data were obtained from SISKOHAT (Integrated Hajj Information System) under the Ministry of Religious Affairs in Malang City, covering the period from 2008 to 2019. 2. Data Preprocessing: > Duplicate entries were removed and missing values handled using techniques documented in missing_values.xlsx. > Data type checks and descriptive statistics were compiled (data_types.xlsx, descriptive_stats.xlsx). > The initial dataset structure is documented in struktur_data_awal.xlsx. > Data were normalized using MinMaxScaler and split into training (80%) and testing (20%) datasets (Preprocessing_Hasil_ManualSplit_80_20.xlsx). 3. Sequence Windowing and Additional Normalization: > A sliding window technique was applied to prepare sequential data for LSTM input, stored in X_Train_80_Windowed_Normalized.xlsx and X_Test_20_Windowed_Normalized.xlsx. 4.Prediction and Evaluation: > LSTM models were trained, and predictions were generated. The file Hasil_Error_LSTM_Tanpa_0.xlsx contains error calculations excluding entries with target = 0. > The final denormalized predictions are saved in hasil_prediksi_lstm_full_denormalized_FINAL.xlsx. > Comparative prediction results between conventional and LSTM methods are stored in hasil_prediksi_lstm_terpisah.xlsx.
Institutions
- Universitas Islam Negeri Maulana Malik Ibrahim Fakultas Sains dan Teknologi