nestauk / afs_neighbourhood_analysis Goto Github PK

Includes:

• Scripts to cluster LADs robustly
• Scripts to benchmark clustering vs. heterogeneity in outcomes

Investigating additional indicators such as the Libraries dataset, the Access to Healthy Assets and Hazards data, Public Transport etc. which may be linked to school readiness. These are datasets not previously used by Nesta.

https://dataforgood.facebook.com/dfg/tools/social-connectedness-index

Definition of done: The robust clustering pipeline used in the industrial taxonomy framework has been refactored to be used with this project.

Change the outcome measure in the clustering pipeline to the percentage of children on FSMs reaching the expected level of development.

Add an alternative outcome measure which is the 2022 EYFSP results. The outcome will be the percentage of children on free school meals reaching a good level of development.

Definition of done: The EYFSP scripts have been transformed into Metaflow pipelines and retrieve the EYFSP data successfully from the DfE website.

Exploring the Fingertips PHE dataset. Focusing on:

• C03c - Smoking in early pregnancy
• C04 - Low birth weight of term babies

Definition of done: Add the datasets for the indicators identified as missing into a Metaflow pipeline.

Looking at the Education statistics from the DfE. In particular looking at:

• Childcare and Early Years: Education provision children under 5
• Early Years Foundation Stage: Early Years Foundation Stage Profile results

• Save formatted charts
• Save choropleths

Look at known causal influencers on EYs development and compare how they change within the clusters and between them.

Definition of done: We have identified which indicators in the Fingertips dataset are to be included in the clustering pipeline.

Using correlation analysis, work out which indicators are most correlated with EYFSP data outcomes. We also need to identify which indicators are strongly correlated with one another as we will only need one of these indicators.

• Fingertips
• DfE data