Skip to main content

Table 2 The logistic regression model’s b coefficients

From: Using mobile money data and call detail records to explore the risks of urban migration in Tanzania

Rank

Feature

Coef.

1

Deprivation of subward most phoned

0.56

2

Average amount (TZS) spent per day

−0.39

3

Home region’s education level∗

−0.22

4

Deprivation of ward most phoned

0.18

5

Home region: Pwani

0.15

6

Percentage of calls to Dar es Salaam

0.14

7

Home region: Mara

0.13

8

Home region: Shinyanga

−0.13

9

Entropy of phone numbers called

−0.12

10

Home region: Manyara

0.11

  1. Note: For the binary outcome variable 1 = Migration to poorer subwards, 0 = Migration to more affluent subwards. TZS = Tanzanian Shillings. Features are ordered by the absolute coefficient value, with the most predictive feature first. All features were calculated on data before the individual migrated to Dar es Salaam. ∗From the 2014 UN Human Development Report for Tanzania. The variable measures the contribution of education to deprivation in the region, where a higher number represents poorer education.