make_neighbouring_data

Description Extracting new covariates from the data using spatial correlation. To extract a new covariate, for each spatial unit, the values of an existing covariate (spatial or temporal) are averaged in the neighboring spatial units of this unit. It can be repeated for several neighbouring layers, where each layer includes the neighbouring spatial units with a certain distance. The first layer contains adjacent neighbours, the second layer contains neighbours with a distance of one spatial unit, and so on for other layers.

Usage

preprocess.make_neighbouring_data(data, column_identifier=None, number_of_layers=1, neighbouring_matrix=None, time_dependency_flag=1, verbose=0)

Parameters

#	Input Name	Input Description
1	data	type: data frame or str default: - details: a data frame or address of a data frame containing temporal or spatial covariates and the id of spatial (and temporal) units. The data must include the following columns with names in the specified format in the description, or if the columns have arbitrary names their content must be specified using the column_identifier argument. Spatial id: The id of the units in the finest spatial scale of input data must be included in the data with the name ‘spatial id level 1’. The spatial id must have unique values. Temporal ids: If data is time-dependent (i.e., time_dependency_flag = 1), the id of the temporal units recorded in the input data must be included in the data with the column name ‘temporal id’. The temporal ids must have unique values and a sortable format. The time dimension of data can also be specified in multiple scales (e.g., hour, min, sec). In that case, each scale must have a separate temporal id with a column name of the format ‘temporal id level x’, where ‘x’ is the corresponding temporal scale level, starting with level 1 for the smallest scale (e.g. sec). The combination of these temporal scale levels’ ids should form a unique identifier. All the remaining columns are considered as covariates except those with names of the format ‘spatial id level x’ or ‘temporal id level x’ where x is the level of the corresponding spatial or temporal scale.
2	column_identifier	type: dict or None default: None details: If the input data column names do not match the specific format of temporal and spatial ids and covariates (i.e., ‘temporal id’, ‘temporal id level x’, ‘spatial id level x’), a dictionary must be passed to specify the content of each column. Keys must be a string in one of the formats: {‘temporal id’, ‘temporal id level x’,’spatial id level x’} The values of ‘temporal id level x’ and ‘spatial id level x’ should be the names of the columns containing the temporal and spatial ids in the x scale level respectively. If the input data has only one temporal scale, the name of the column including the temporal ids must be specified with the key ‘temporal id’. example: {‘temporal id level 1’: ‘week’,’temporal id level 2’: ‘year’,’spatial id level 1’: ‘county_fips’, ‘spatial id level 2’: ‘state_fips’}
3	number_of_layers	type: int default: 0 details: The number of neighbouring layers Each neighbouring layer for a spatial unit includes neighbours with a certain distance from this spatial unit. The first layer contains adjacent neighbours, the second layer contains neighbours with a distance of one spatial unit, and so on for other layers. For each covariate, the average values of this covariate in the spatial units included in a neighbouring layer are added to the data as a new covariate. Therefore, neighbouring_layers = n, for each covariate, adds n new covariates to the data.
4	neighbouring_matrix	type: numpy.ndarray or None default: None details: The adjacency matrix of spatial units A two-dimensional binary array with dimensions equal to the number of spatial units included in the data. The value of each element of this array indicates the adjacency (value 1) or non-adjacency (value 0) of two spatial units. Note that the order considered for the spatial units in the rows and columns of the matrix should be based on their spatial id order (numerical order for numeric ids or lexicographical order for string ids). example: numpy.array([[0,1,0],[1,0,1],[0,1,0]])
5	time_dependency_flag	type: int default: 0 details: Time dependency of the data 0 if the data is time-dependent or 1 otherwise. In the second case, data should be a single data frame or its address.
6	verbose	type: int default: 0 details: The level of details in produced logging information available options: 0: no logging 1: only important information logging

Returns

#	Output Name	Output Description
1	neighbouring_data	type: dataframe details: a data frame including extracted covariates from the neighbourhood. The name of the extracted covariates have a special format, e.g. the covariate obtained by averaging the ‘temperature’ in spatial units of the first neighbouring layer has the name ‘temperature_l1’.

Example

import pandas as pd
from stpredict.preprocess import make_neighbouring_data

df = pd.read_csv('USA COVID-19 temporal data.csv')


neighbouring_data = make_neighbouring_data(data = df, number_of_layers = 2,
                                             neighbouring_matrix = [[0,1,...,0],
                                                                    [1,0,...,1],
                                                                    ...,
                                                                    [0,1,...,0]])