WEBVTT 00:00.700 --> 00:06.340 In this video we will learn how to use the shift method and actually this method is a very powerful 00:06.340 --> 00:13.240 method that allows us to calculate the absolute change also the relative change in stock price from 00:13.240 --> 00:18.160 yesterday to today or from last week to today or whatever. 00:18.160 --> 00:19.590 So we are still important. 00:19.600 --> 00:21.940 Our close data frame. 00:22.150 --> 00:28.480 So here we have the closing prices for the six stocks and in this video we want to only work with the 00:28.480 --> 00:29.620 Apple stock. 00:29.620 --> 00:31.420 So we selected the apple column. 00:31.420 --> 00:38.350 We create a copier and we transform the panel serious into a data frame and then we store the new data 00:38.350 --> 00:41.520 frame the apple data frame in the available Apple. 00:42.970 --> 00:45.160 And let's have a look here. 00:45.190 --> 00:52.520 So this is our apple data frame with daily stock prices from 2009 until 2019. 00:52.630 --> 00:58.660 And let's assume that for each day we want to calculate the price increase or decrease from the last 00:58.660 --> 00:59.890 trading day. 00:59.890 --> 01:07.820 So for example on the fourth of January the increase from the last trading day was about or point four 01:07.870 --> 01:13.600 dollars and instead of calculating the absolute difference we could also calculate here the relative 01:13.600 --> 01:19.390 difference or the percentage change and therefore on very fast that we want to shift down all values 01:19.390 --> 01:23.170 here in the applet column by actually 1 timestamp. 01:23.440 --> 01:29.740 So we want to have here the price at the thirty first of December we want to have this now at the 4th 01:29.740 --> 01:30.420 of January. 01:30.430 --> 01:36.580 So we want to shift that by one timestamp and we can do this with the shift method. 01:36.580 --> 01:38.950 So let's have a look here. 01:39.010 --> 01:45.970 So the shift method actually shifts the index or we could also say the value is so it's relatively they 01:45.970 --> 01:47.020 are shifted. 01:47.020 --> 01:53.890 So the shift method shifts the index by a desired number of periods and we have fear the pyramid up 01:53.970 --> 01:56.610 periods and by default it set to 1. 01:56.650 --> 02:00.870 But we can also set here different numbers of course. 02:00.880 --> 02:09.420 So now let's try here a shift of 1 so he we can see at the very first timestamp that we have now our 02:09.410 --> 02:12.750 missing value because there is no prior value here. 02:12.860 --> 02:20.530 Then on the fourth of January we have the former value of the thirty first of December so thirty point 02:20.570 --> 02:21.270 one zero. 02:21.290 --> 02:25.160 And so on and of course we can also shift by more than one period. 02:25.180 --> 02:28.290 So let's pass here too. 02:28.550 --> 02:34.040 So now we can see here that we have here two missing values and the very first value we have yet the 02:34.040 --> 02:41.510 first of January and this is the value from the thirty first of December which is actually two business 02:41.510 --> 02:44.150 days or trading days here of a. 02:44.150 --> 02:50.740 And instead of passing a positive number so 1 or 2 you can also pass a negative in the test. 02:50.780 --> 02:53.590 So for example minus one. 02:53.870 --> 02:59.180 And by doing so we are actually a relatively a shifting up here the values in the column. 02:59.210 --> 03:05.480 So now here for example for the thirty first of December we have the value that was formerly here at 03:05.900 --> 03:07.740 the 4th of January. 03:08.060 --> 03:13.620 So for each timestamp here and our shift that data frame we have an hour the value of the next stop 03:13.620 --> 03:20.420 for the following business day and in some cases that can make sense but usually we are shifting in 03:20.570 --> 03:22.070 the other direction here. 03:26.570 --> 03:33.400 So for each timestamp we fear the values from the past and we can compare current values with past values. 03:33.400 --> 03:37.250 So this is a frequent workflow and Finance for example. 03:37.640 --> 03:45.050 And once we are happy with our shift then we can also create a new column and save the sifted values. 03:45.050 --> 03:48.410 So for example we can call you at the column like one. 03:48.410 --> 03:55.960 So we have a lack of one period here and lets the re inspect our data frame. 03:56.100 --> 04:02.270 So are on the right hand side we have now our like one and now we can actually calculate the daily increase 04:02.270 --> 04:10.940 or decrease in stock prices and we can do this by selecting the upper column and subtract actually our 04:10.940 --> 04:14.780 like one column and we can do this here with the sub method. 04:14.810 --> 04:15.890 So let's do this year 04:19.240 --> 04:25.030 and see we can see for example that the EPA's stock price increased from the study first of December 04:25.030 --> 04:30.220 to the fourth of an area by or point forty six dollars. 04:30.340 --> 04:33.760 So this is exactly here this difference. 04:33.790 --> 04:35.020 All right. 04:35.170 --> 04:40.630 Now we can also create a new column called the Friends where we actually store here the pandas series 04:40.840 --> 04:42.340 that we created here. 04:42.700 --> 04:47.890 So it's the daily increase or decrease in stock price. 04:47.890 --> 04:50.030 So now we have you on the right hand side. 04:50.050 --> 04:52.720 The difference between those two columns. 04:53.410 --> 04:59.260 So here we have now the absolute difference but in finance typically it's more meaningful to have a 04:59.260 --> 05:04.030 relative increase or decrease or a percentage change are in finance terms. 05:04.030 --> 05:12.010 Also call it return and we can calculate the percentage change or return by selecting the above column 05:12.430 --> 05:19.580 and then we divide here the values in the upper column by the values in the like one column. 05:19.690 --> 05:21.550 So let's do this here 05:24.850 --> 05:30.880 and for example here we can see that from the thirty first of December to the 4th of January the stock 05:30.940 --> 05:34.170 increased by one point fifty five percent. 05:34.420 --> 05:40.390 So if you want to get now the percentage change or the stock return in percentage terms to be of all 05:40.390 --> 05:43.830 have to subtract one here so we can do this. 05:43.830 --> 05:49.750 This the sub method and then we have to multiply its value by 100 05:52.340 --> 05:56.930 so give us now the daily percentage changes our returns. 05:56.930 --> 06:02.140 So for example for the Fourth of 10 year weaken CBS a return of one point five five percent. 06:02.870 --> 06:10.420 And for example on the 6th of January we have a negative return off from minus the one point five nine 06:10.430 --> 06:11.650 percent. 06:11.660 --> 06:19.640 All right and finally we can also create a new column percentage change where we actually pass uh the 06:19.640 --> 06:21.380 various independent serious 06:24.080 --> 06:27.170 and now we can finally inspect our data frame. 06:27.710 --> 06:34.670 So here in this video we created a like one column where we actually sift at the daily prices by one 06:34.760 --> 06:42.110 timestamp and by doing so we could calculate the daily increase or decrease in stock prices and we could 06:42.110 --> 06:45.790 also calculate the daily percentage change return. 06:45.980 --> 06:51.130 And with this we are finished with this video and I hope to see you also in the next one by.