The context of the problem that I am dealing with is trying to convert the results from a time series forecast, plotted with matplotlib.plotly back into a dataframe so that I can use the cufflinks library to be able to get a more interactive chart going so that I can hover over data points to get a more detailed look at the forecast.
so after training and creating a simulation the code goes:
date_ori = pd.to_datetime(df.iloc[:, 0]).tolist()
for i in range(test_size):
date_ori.append(date_ori[-1] + timedelta(days = 1))
date_ori = pd.Series(date_ori).dt.strftime(date_format = '%Y-%m-%d').tolist()
date_ori[-5:]
accepted_results = []
for r in results:
if (np.array(r[-test_size:]) < np.min(df['Close'])).sum() == 0 and \
(np.array(r[-test_size:]) > np.max(df['Close']) * 2).sum() == 0:
accepted_results.append(r)
len(accepted_results)
accuracies = [calculate_accuracy(df['Close'].values, r[:-test_size]) for r in accepted_results]
plt.figure(figsize = (15, 5))
for no, r in enumerate(accepted_results):
plt.plot(r, label = 'forecast %d'%(no + 1))
plt.plot(df['Close'], label = 'true trend', c = 'black')
plt.legend()
plt.title('average accuracy: %.4f'%(np.mean(accuracies)))
x_range_future = np.arange(len(results[0]))
plt.xticks(x_range_future[::30], date_ori[::30])
plt.show()
I have started to dissect the last plotting section to attempt to convert the data into a dataframe in order to plot with cufflinks as the format for cufflinks is like :
import cufflinks as cf
# data from FXCM Forex Capital Markets Ltd.
raw = pd.read_csv('http://hilpisch.com/fxcm_eur_usd_eod_data.csv',
index_col=0, parse_dates=True)
quotes = raw[['AskOpen', 'AskHigh', 'AskLow', 'AskClose']]
quotes = quotes.iloc[-60:]
quotes.tail()
AskOpen AskHigh AskLow AskClose
2017-12-25 22:00:00 1.18667 1.18791 1.18467 1.18587
2017-12-26 22:00:00 1.18587 1.19104 1.18552 1.18885
2017-12-27 22:00:00 1.18885 1.19592 1.18885 1.19426
2017-12-28 22:00:00 1.19426 1.20256 1.19369 1.20092
2017-12-31 22:00:00 1.20092 1.20144 1.19994 1.20147
qf = cf.QuantFig(
quotes,
title='EUR/USD Exchange Rate',
legend='top',
name='EUR/USD'
)
qf.iplot()
Where I have gotten so far is trying to dissect the plotly graph into a dataframe as so, these are the forecasted results:
df = accepted_results
rd = pd.DataFrame(df)
rd.T
0 1 2 3 4 5 6 7
0 768.699985 768.699985 768.699985 768.699985 768.699985 768.699985 768.699985 768.699985
1 775.319656 775.891012 772.283885 737.763376 773.811344 785.021571 770.438252 770.464180
2 772.387081 787.562968 764.858772 737.837558 775.712162 770.660990 768.103724 770.786379
3 786.316425 779.248516 765.839603 760.195678 783.410054 789.610540 765.924561 773.466415
4 796.039144 803.113903 790.219174 770.508252 795.110376 793.371152 774.331197 786.772606
... ... ... ... ... ... ... ... ...
277 1042.788063 977.462670 1057.189696 1262.203613 1057.900621 1042.329811 1053.378352 1171.416597
278 1026.857102 975.473725 1061.585063 1307.540754 1061.490772 1049.696547 1054.122795 1117.779434
279 1029.388746 977.097765 1069.265953 1192.250498 1064.540056 1049.169295 1045.126807 1242.474584
280 1030.373147 983.650686 1070.628785 1103.139889 1053.571269 1030.669091 1047.641127 1168.965372
281 1023.118504 984.660763 1071.661590 1068.445156 1080.461617 1035.736879 1035.599867 1231.714340
then converting the x axis from
plt.xticks(x_range_future[::30], date_ori[::30])
to
df1 = pd.DataFrame((x_range_future[::30], date_ori[::30]))
df1.T
0 1
0 0 2016-11-02
1 30 2016-12-15
2 60 2017-01-31
3 90 2017-03-15
4 120 2017-04-27
5 150 2017-06-09
6 180 2017-07-24
7 210 2017-09-05
8 240 2017-10-17
9 270 2017-11-20
lastly I have the close column and this is what I've been able to come up with for it so far
len(df['Close'].values)
252
when i use
df['Close'].values
I get an array, I'm having problems getting this all together, the cufflinks iplot graphs are just way better, and it would be amazing if I could somehow gain the intuition to do this, I apologize in advance if I didn't try hard enough, but I'm doing my best I can't seem to find the answer no matter how many times I've searched google so I thought I would ask here.
This is what I did, I went through and printed indipendent strings like print(date_ori) as well as simplified it with print(len(date_ori) which in turn had all of the dates for the forecast, then i made it into a dataframe with df['date'] = pd.DataFrame(date_ori), where as with the results, I had to transpose them with df.T so they would be in a long column format rather than in a long row, so first
then
I had trouble naming the column 0 which contained all of the predicted results so i just saved the file with
then i edited the column named 0 to results and added .csv to the end, then pulled the data back into memory
then i formatted the date
and dropped the un needed columns, and i guess i could add the close data that i started with to plot together now, but i got the results into the dataframe so now i can use these awesome charts! Can't believe i figured it out within 18 hours I was so lost lol.
also i dropped the experiment to just one simulation so there was only 1 row of results to deal with so i could figure it out.