datetime format

[qn222ad]

The strange thing was that if I used data from yahoo finance, the add_line function worked very good. chart plot dates nice and clear. However, yahoo finance did not had all the data I want then I need to use data from csv file.
As I used data from cvs file, then add_line did not work.
I used the date format as follows: dates= pd.to_datetime(pivot_e.index.date).astype('int64')
What is the potential problem of this?
Thanks :)

[highfestiva]

Don't cast do int64 unless you have to.

[qn222ad]

data from yahoo finace:

data from csv:

In data from csv file, I set both index in data file and csv file of dates using pd.to_datetime(dfn['Date'])
dfn.loc[:,'Date'] = pd.to_datetime(dfn['Date'], errors='coerce', utc=True)
dfn = dfn.set_index('Date')
date time show in pivot dates:

and in data:

I changed both index to second, but still the same... did not know where the problem.
Please advise if you know.
Thanks

[qn222ad]

This is example code:

import yfinance as yf
import pandas as pd
pd.set_option('display.max_colwidth', None)
import finplot as fplt

selected = []
def read_days_only(df):
    dfn = df.copy()
    dfn.loc[:,'Date'] = pd.to_datetime(dfn['Date'], errors='coerce')
    dfn = dfn.set_index('Date')
    dfn['Date'] = dfn.index
    return dfn
def extract_price(df1,df2):
    df1 = df1[pd.to_datetime(df1.index.date.astype(str)).isin(pd.to_datetime(df2.index.date))]
    return df1

pivot = read_days_only(pd.read_csv('extracted.csv'))

data = read_days_only(pd.read_csv('data/BankNifty.csv'))
#data from yahoo fianance
#data = pd.DataFrame(yf.download("^NSEBANK",period='max'))

data['Date'] = data.index
data['Date'] = pd.to_datetime(data['Date']).astype('int64')
data['High']=data['High'].values+4 #old data dont have 
data['Low']=data['Low'].values-4 #old data dont have 
mx=2*max(data.High)
pivot_e=extract_price(data,pivot)
dates= pd.to_datetime(pivot_e.index.date)
fplt.candlestick_ochl(data[['Open','Close','High','Low']])
for x, (d, price) in enumerate(zip(dates, pivot_e.Close)):
    fplt.add_line((d, 0), (d, mx), color='#CC9EC1', style='--')
fplt.show()

[highfestiva]

Send all files in a zip. You're handling the index wrong. You probably want something like this:

data = data.reset_index() # gets you the index into a column
...
data = data.set_index('my_date_column') # set the index back to using datetime again

This has nothing to do with finplot, but pandas. It takes some time to learn the basics, but I suggest you do. GL!

[qn222ad]

Yes, I am new to coding:D
I checked data type of both
data = read_days_only(pd.read_csv('data/BankNifty.csv'))
print(data.index.inferred_type == "datetime64")
data = pd.DataFrame(yf.download("^NSEBANK",period='max'))
print(data.index.inferred_type == "datetime64")
both return True
then they must be at the same type
I will read about datetime in pandas. However, I did not find any different between the two :(

[highfestiva]

I ran your example, it shows dates. What is the problem?

[qn222ad]

I can plot them now,
Thank you very much!
I need to change datetime structure (origin dd/mm/yyyy) need to replace / to - then it work

[highfestiva]

datetime64[ns] is fine. You're probably doing something like plot(df['X']) without having datetime in the index. See the tutorial in the wiki.

[qn222ad]

I always had datetime in index, because i used the index date to link with other data table. You see two pictures above, all was in index. The datetime format take me alot of time, the thing is I do not know where to find error because its format was datetime64[ns] and the same

[highfestiva]

The second one has three of each period. That does not constitute a time series. A time series is a bunch of data in sequence.

[qn222ad]

Thank for good spot, it was error from using pd.merge. However, After fix this, I still had the same issue:
this is my input for line

The data is not in sequence:

and the result:

Where did it goes wrong?
print(data.index.inferred_type == "datetime64")
print(pivot.index.inferred_type == "datetime64")

result
True
True

[highfestiva]

So sort it then.