Query interface examples

[schmitts]

Could you please add examples on how to query/select? E.g. how does a pandas query like below look in DataFrame?

df_filtered = df[df.something > 5]

[hosseinmoein]

The equivalent to that, in DataFrame, is either of the following methods:

get_data_by_sel()
get_view_by_sel()

First one returns another DataFrame. Second one returns a view

The documentation for them is here:
https://htmlpreview.github.io/?https://github.com/hosseinmoein/DataFrame/blob/master/docs/HTML/get_data_by_sel.html

There are examples in the tester file here:
https://github.com/hosseinmoein/DataFrame/blob/master/test/dataframe_tester_2.cc#L4396
and here:
https://github.com/hosseinmoein/DataFrame/blob/master/test/dataframe_tester_2.cc#L4621

There are similar examples to your questions in the tester file

[yegorrr]

how can i do something like
df["price_usd"]=df["price"]*df["fx_rate"]
or
df["x"]+=constant
for each timestamp(including missing data), ie something like a visitor but for rows instead of columns

[hosseinmoein]

There are a couple of different ways:
Look at consolidate()
You can write your own visitor which is simple
Or you can manually do it with a loop

[yegorrr]

thanks, consolidate is probably what i was looking for.

another question, assume i'm adding last trade pices of some asset in real time and i'm only interested in its current price and moving average over x seconds. what would be the most efficient way:

1. store all trades and timestamps as they arrive in data frame and on each insert call get_data_by_sel() filtering all trades over last x seconds and averaging that data
2. create a separate structure, kind of a queue that stores only data over the last x seconds - each insert may pop an arbitrary amount of elements from the queue - from 0 to N. assuming all data is sorted by timestamps, this would be as efficient as 1)
3. something else?

[hosseinmoein]

If you don’t need all the data, why store it? In that case something like 2 is better