Temporal Disaggregation Of Time Series In Python

I am trying to find a package that enables temporal disaggregation of timeseries. There is a package in R called tempdisagg. https://journal.r-project.org/archive/2013/RJ-2013-028

Solution 1:

I've created an open source Python package called timedisagg that is based on the R tempdisagg package. The package implements the basic Chow-Lin and Litterman methods. It also allows for basic average, sum, first and last conversion choices like the R package.

Given the following function call in R to disaggregate sales.a as a function of exports.q:

model <- td(sales.a ~ 0 + exports.q,method="chow-lin-maxlog",conversion="sum")

A similar call can be made using timedisagg as below:

from timedisagg.td importTempDisaggtd_obj= TempDisagg(conversion="sum", method="chow-lin-maxlog")
final_disaggregated_output = td_obj(expected_dataset)

where the expected_dataset is a pandas dataframe with the following format:

indexgrainXy01972      11432.63900          NaN11972      21456.89100          NaN21972      31342.56200          NaN31972      41539.39400          NaN41973      11535.75400          NaN51973      21578.45800          NaN61973      31574.72400          NaN71973      41652.17100          NaN81974      12047.83400          NaN91974      22117.97100          NaN101974      31925.92600          NaN111974      41798.19000          NaN121975      11818.81700   136.702329131975      21808.22500   136.702329141975      31649.20600   136.702329151975      41799.66500   136.702329161976      11985.75300   151.056074171976      22064.66300   151.056074181976      31856.38700   151.056074191976      41919.08700   151.056074..............1522010      119915.79514988.3096761532010      219482.48000988.3096761542010      318484.64900988.3096761552010      418026.46869988.3096761562011      119687.52100NaN1572011      218913.06608NaN

Here X is is exports.q and y is sales.a.

The output final_disaggregated_output will appear as below where y_hat is the disaggregated sales:

indexgrainXyy_hat01972      11432.639 NaN21.65687911972      21456.891 NaN22.21973721972      31342.562 NaN20.85541331972      41539.394 NaN23.93791641973      11535.754 NaN24.229008

Edit - If someone needs help working their data into my package, feel free to raise an issue at the git for the package.