• The simulation generator is list in the next section.
• There are 4 fields, which are latitude, longitude, timestamp, and Grid Index (Arm).
• This directories include all the simulated datasets, that is presented in the manuscript.

The following picture presents one of the example.

• The raw dataset is downloaded from http://www.houstontx.gov/311/.
• It contains all 311 request during the 311 Hurricane strike.

• The detail of the protocol to clean the dataset is elaborated in the manuscript.
• There are 4 fields, which are latitude, longitude, timestamp, and Grid Index (Arm)

• Code is implemented in MATLAB.

1. pois.m contains the function that draw samples from Possion Distribution
2. Hawkes_Simulation.m contains the function that run the simulation given the parameter.

• T is the total simulation time.
• η is the rate per second and T is the total time span.
• φ governs the average number of offspring that an event spawns.
• ω controls the waiting time between parent and offspring.
• σ controls the spatial distance between parent and offspring event

• times: Timestamps
• x: Simulated longitude.
• y: Simulated latitude.
• offspring: A index to show whether it is offsprings or parents.
• cluster_Id: It is the index to indicate which cluster it belongs to.

1. CommandLineParamatize.py: This is the code to read in parameters
2. DataInput.py: This is the code to read in the input data file.
3. MAB.py: Main code to run multi-armed bandit algorithm.
4. HpSpUCB.py: The function that excecute the proposed model, HpSpUCB.

• UnbDataPath: Specify the input data path with four fields, Time, latitude, longitude, and grid index.
• ResultsPath: This is to keep track of every pull results.
• Method: To specify the method.
• NumArm: The number of arms in total.
• NumPulledArm: The number of arms that we pull each visit.
• PeriodPerPull: The period between each pull.

• Gamma: The weight betwee the proposed Hawkes processes component and the exsisted component.
• Lambda: The standward deviation used in Gaussian process and Gaussian filter.
• Delta: The standward deviation used in the Metropolis-Hastings sampling method.
• Tau: The temporature for softmax function.
• AlphaR: The weight on the upper confidence bound on rewards.
• AlphaH: The weight on the uppper conficence bound on event intensities.
• nsample: The number of samples to draw from posterior distribution.
• SimTime: Simulation time for diffrent initial positions.