tags:

views:

131

answers:

0
Q: 

Improving the accuracy of Microsoft Time Series Algorithm

Situation is that we have branches in every city, selling food.

I feed the time series algorithm with the actual date, as the key time, and total sales of that day, as the input and predict. Predictions are not bad. But,

I would like to know if i can improve the predictions by for example feeding with the number of branches(a new branch can increase the sales number of course) or the number of food types?

And If i can, how? If i can't, is there a better approach/algorithm?

related questions

storing massive ordered time series data in bigtable derivatives
Time series in R
Downsize Growing Time-Related Mysql Table
Finding similarities in a multidimensional array
How to use aryule() in Matlab to extend a number series?
lag in apply statement doesn't work in R
which is the most efficient XML Parser for C++ ?
On the issue of automatic time series fitting using R
Changing annotations in time series plots in R
What is the best open source solution for storing time series data?
dimensional and unit analysis in SQL database
Reshaping time series data from wide to tall format (for plotting)
Cumulative sums, moving averages, and SQL "group by" equivalents in R
JFreechart Rendering a timeseries on a tick scale
JFreechart Realtime Combined Plot - Rendering the previous value for a step subchart if no data point has been received
Reconstructing state from time series data events
Probability time series, observed data probabilities (deja vu)
Time-series and correlation strategies
Java library to do time series analysis
rrd tool alternative for high volume
Linear regression for time series with Gnuplot
"Join" of time series
What's a good way to store raster data?
What are some compact algorithms for generating interesting time series data?