wiki.raudys.com

This is an old revision of the document!

---

allign data in DataFrame

#allign data to dataframe
df1 = pd.DataFrame(0.0, index=all_date, columns=[a.sidb for a in d])
start = time.time()
for q in d:
    df1.iloc[np.in1d(all_date, q.date, assume_unique=True), df1.columns.get_loc(q.sidb)] = q.pnl
print ("Result: %.2f secs" % (time.time() - start))

allign data in numpy array (3x faster) # allign using numpy start = time.time() a1 = np.zeros¹⁾) i = 0 for q in d:

  idx = np.in1d(all_date, q.date, assume_unique=True)
  a1[i, idx] = q.pnl
  i = i + 1

print (“Result: %.2f secs” % (time.time() - start))

you can join/merge objects if they have same type axis

for example align two time series by date