A Neural Network model can predict the success of crowdfunding projects
This project is using the Kickstarter (latest 2019) and Indiegogo (latest 2017) datasets collected from Webrobots.io.
This example code can encode kickstarter dataset, which contains numerical, categorical, and textual features, and it's using pre-trained GloVe to encode the textual feature.
export DATA_DIR=$HOME/projects/Predict-The-Success-of-Crowdfunding/raw_data_webrobots
export GLOVE_DIR=$HOME/projects/Predict-The-Success-of-Crowdfunding/glove
export DATA_NAME=KICK
export OUTPUT_DIR=$HOME/projects/Predict-The-Success-of-Crowdfunding/outputs_test
python encoder.py --data_dir $DATA_DIR \
--data_name $DATA_NAME \
--output_dir $OUTPUT_DIR\
--metadata_file metadata_comb.json \
--use_text_features True \
--encode_text_with glove\
--glove_file $GLOVE_DIR/glove.6B.50d.txt \
--max_words 20\
--max_sequence_length 5\
--embedding_dim 50