WEBVTT 00:00.000 --> 00:06.600 Hello and welcome to future. In this video tutorial I would like to show you 00:06.600 --> 00:11.600 how you can create time series from your CSV files. For this we use the CHECK-IN in 00:11.600 --> 00:18.800 our frontend NOW. First, log in at now.future-forecasting.de. Then, in the 00:18.800 --> 00:25.920 menu on the left, select “Data” and choose your dataset. In our case it's the demo data. If you 00:25.920 --> 00:30.820 want to upload a new dataset, just click the plus, drag-and-drop the data here 00:30.820 --> 00:36.460 or select them after clicking the plus. Then, for the CSV file that 00:36.460 --> 00:42.360 contains your data, click the “CHECK-IN” button and you will be taken to the next step. We see 00:42.360 --> 00:48.660 here a preview of our data. We see the different columns and values. In our case 00:48.660 --> 00:54.040 we want to delete the “Region” column because we won’t need it later on. We 00:54.040 --> 00:58.200 simply click on the column and confirm the deletion by pressing the button. 00:58.920 --> 01:04.660 We now proceed to the next step. Here we define the structure of the CSV file a bit 01:04.660 --> 01:09.160 more precisely. First, from a list of values we can specify the decimal separator or the 01:09.160 --> 01:14.140 thousands separator. In addition, we can provide strings that should be interpreted as missing values 01:14.140 --> 01:19.080 Next, we specify the column that contains the date information 01:19.080 --> 01:24.000 and choose the appropriate date format from a list of values. 01:25.920 --> 01:33.120 We indicate that the “Demand” column contains our values and that the columns “Material” and 01:33.120 --> 01:38.140 “Customer” hold grouping values that will later be assigned as hierarchical information to the time series 01:38.140 --> 01:43.700 We now continue to the next step. In this step we can 01:43.700 --> 01:49.420 check the preparation of the raw data. On the left we see whether there are problems with the data. We can 01:49.420 --> 01:56.980 fix these by clicking on the exclamation mark. We see that our date column contains an invalid date format 01:56.680 --> 02:02.620 and want to remove this row. We see that there are two different problems in our Demand column: 02:02.920 --> 02:10.120 There are both missing values and empty values. In both cases we want to remove the corresponding rows. 02:10.120 --> 02:14.000 On the right we see various visualizations of our data: 02:14.000 --> 02:22.720 we see the number of data points per day, we see the distribution of the values, and we see the distribution across the individual groups. 02:23.680 --> 02:38.880 We now proceed to the final step. Because we want to create time series at a monthly level, we select “monthly” as the temporal aggregation. We could also set a start and end date for the time series here. 02:38.880 --> 02:42.000 We want to create time series for the materials, so we select “Material” as the data level. 02:42.300 --> 02:52.880 There is one material for which we do not want to create a time series; accordingly, we create an exclusion filter for this material. We select it from the dropdown list. 02:53.680 --> 02:58.680 As the variable we select “Demand” because this column contains our values. 02:58.680 --> 03:05.620 Individual data points should be aggregated as a sum, and missing data points should be filled with 0 in the target time series. 03:05.620 --> 03:11.640 We click on “Prepare time series” and wait a brief moment. 03:11.640 --> 03:20.680 We now see the finished, processed time series. In the table we can select individual materials to view them individually. 03:20.680 --> 03:26.080 We can either download the results as a CSV or create forecasts for them. 03:26.640 --> 03:29.220 More about that in the next video.