Python/Pandas Multi Time Frame handling?

how do you guys store and process your mutli time frame data?
I have 1 minute data indexed by time stored in hdf5.

o h l c
b'dt'
2017-01-29 23:01:00 1.0711 1.0711 1.0710 1.0710

I was wondering how do you store for example 1 hour or daily data (resample('1440T')) and then for example calculate MAs or ATR on higher time frame data?

I have recently started testing some ideas with Python/Panda in order to try some AI /deep machine learning stuff. I think maybe after resampling you could store the data as new HDF5 to make things faster and easier to test on longer data. I haven't tried to create HDF5 for very long data , say 10 years or so at tick level data , so not sure how the performance would be .Though you could do a select() for the time space which you will plot or analyze, (i.e. a given month) , tean it will probably load fast.

For MA & ATR, you could add new columns (calculated) after resampling.

Rolling() should give moving average
ewm() for exponential moving average
ATR can be calculated by using the absolute (abs) of high-low difference

Im going down that route too
are you going to use Keras for AI?

yea Keras/TensorFlow . I am planning to add news as input (with forecast data at least) as well. still looking for historic news data available for download.

News data sounds complicated.
Can you not scrape historic news from FF?
I got 15 years of 1m data from http://www.forextester.com/data/datasources
what kind of data you have?

Very easy. I have a class that calculate this and implemented all standard indicators usin that class