README.md

In DFE-Capital/Blockbuster-2: Deterioration Model of the School Estate

This package represents the most current version of the Blockbuster Deterioration model. Code optimisation resulted in a completely rewritten code base so it is appropriate to initiate a new github page here. The original version can be found at: (https://github.com/DFE-Capital/Blockbuster).

Government related plot themes and palettes are from (https://github.com/ukgovdatascience/govstyle).

blockbuster2 - The Blockbuster Deterioration Model

This is a prototype package being developed in anticipation of analysing the Condition Data Collection and is subject to constant development

This R package allows you to simulate the deterioration of School buildings through time using a Discrete Time Markov Chain. R packages are an ideal way to package and distribute R code and data for re-use by others.

Data on the condition of the school estate were collected during the Property Data Survey Programme (PDSP) of 2012-2014. Approximately 2.7 million rows of data were collected. This provides the initial state of the School Estate at timestep zero. The deterioration of the School Estate is then modelled by using deterioration rates associated with each construction element / sub-element / construction-type. Cost of repairs for these building components are estimated. An investment profile for rebuilding and maintaining the School Estate can also be input to mitigate this entropic spiral.

The PDS data is not included with this package. A small simulated dataset is provided.

The model works with the following parameters

Data inputs Description

Input data for time zero A data frame with one row per construction element

Deterioration rates Estimated probabilities of deterioration within one year for each component at each condition grade

Repair costs Repair costs of repairing components at condition grades B, C, D, and E

Rebuild costs Cost in pounds per square metre used to estimate cost of rebuilding individual buildings

Repair funds A vector containing the budget for repairs each year

Rebuild funds A vector containing the budget for rebuilding each year

Inflation A vector containing the inflation to be applied to all repair and rebuild costs each year

## Installing the package

The package can be installed with the devtools package with devtools::install_github('DFE-Capital/blockbuster2').

If you cannot use this function (due to firewalls for instance) you can download the package as a .zip file from the main repository page, and run devtools::install_local('path_to_zip_file'). You may need to install Rtools to facilitate this.

## Using the package

## From an R session

Launch an R session as normal and run the following (again setting the arguments as required):

library(blockbuster2)

The beauty of package development is that the business knowledge is enshrined in the package, with the code and documentation in one place. To familiarise yourself after install you can work through the blockbuster vignette which will provide typical uses. The blockbuster-methodology vignette provides in depth detail about how the model works.

You can also use the extensive R help which explains the input arguments and output values of all functions by prefixing a function with a ?.

?Blockbuster

## Example

The following code will run the Blockbuster model on the simulated data over a 10 year period and store the output in an object called results:

results <- Blockbuster(simulated_elements, forecast.horizon = 10) Note that by default the Blockbuster Deterioration Model automatically saves yearly states into files called blockbuster_output_element_*.rds and blockbuster_output_block_*.rdswithin an output sub-folder of the current working directory that the function will create if it does not already exist. The complete output is saved as blockbuster_output_output.rds. This behaviour is customisable. Saving of outputs can be turned off with save = FALSE passed as an argument. Alternatively, the path and filename can be adjusted using the filelabel and path arguments. When using large datasets and forecasting over long periods it is recommended to always use save = TRUE, to avoid losing progress should R run out of memory to hold the complete output.

The element level output is a copy of the input, but with updated grade probabilities and repair totals. Note that the the cost of repairing an element at a given grade is given by the *.repair.total column. The *.repair.cost column is the unit cost of repairing that component type.

We recommend using the functions pull_Element_Data, pull_Block_Data, load_Element_Data and load_Block_Data to put the output of a Blockbuster Deterioration Model run into a friendlier format for analysis.

Passing numeric vectors to the rebuild.money and repair.money arguments allows control of the amount of money available for rebuilding and repairing each year respectively.

## Project goal

To simulate the deterioration of the School Estate into the future under various maintenance and rebuilding spending policies. This will provide a data driven approach to improving the management of the School Estate potenitally saving public money. The model's predictions should be evaluated with the next Condition Data Collection due in 2017 and beyond. This package will also provide the skills and learnings necessary to extend this package functionality for the CDC data.

Improvements

• If you find any bugs raise an issue on the Github package page.
• Any suggestions or improvments to the decision making process within the model would also be welcome.
• Package development is iterative and dependent on user feedback.

DFE-Capital/Blockbuster-2 documentation built on May 26, 2019, 7:23 a.m.