razib764/hip-tensorflow

Hawkes Intensity Process Implemented in TensorFlow

hip-tensorflow

Hawkes Intensity Process Inspired by Expecting to be HIP: Hawkes Intensity Processes for Social Media Popularity Implemented in TensorFlow.

Installation

Run the following snippet to install hip in a Mac/Unix environment.

git clone https://github.com/kasrsf/hip-tensorflow.git
cd hip-tensorflow
pip install -e . #install in editable mode  

To verify successful installation, the following command should run from the command line without any errors:

python -c "import hip"

Usage

Single Feature Analysis

To run the feature analysis with the hawkes model, run the following command:

python hip_single_feature_analysis.py [input_dir] [feature_index]

the script will train the HIP model using all the csv files in input_dir with a single exogenous source feature_index and output a tab-seperated table containing the learned values.

Example

> python hip_single_feature_analysis.py data/sample_data_2 2

eta    mu    theta
0.20671242    1.3099899    13.408139

Requirements

• Python 3.5

The code may work with other versions of Python but this is not tested.