Technology Blogs by SAP
Learn how to extend and personalize SAP applications. Follow the SAP technology blog for insights into SAP BTP, ABAP, SAP Analytics Cloud, SAP HANA, and more.
cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 

Quickly load Covid-19 data with hana_ml and see with DBeaver

Vitaliy-R
Developer Advocate
Developer Advocate
‎03-11-2020 10:49 PM
Well, analyzing the spread of the SARS-CoV-2 coronavirus was not my dream use case. But based on the responses to ferrygun18's Tracking Coronavirus COVID-19 Near Real Time with SAP HANA XSA article I decided to add my two groszy too.

[Updated on 20-03-30 with the changed links to the source data; and the new map output based on the new data granularity. Thanks douglas.maltby2 for your comment!]

In his blog post, Ferry used JavaScript in SAP HANA XSA to pull the data from CSV files updated daily by Johns Hopkins University.

I would like to show you how you can pull and load these files into SAP HANA using just a few lines of code thanks to SAP HANA Python Client API for Machine Learning (hana_ml package).
Some people were confused with the visualization on the map at the end -- please note that this article focuses on technical use case connecting different components, not on doing coronavirus data deep analysis.

Get Python environment, e.g. Jupyter


I will use Jupyter in the Docker container for that. Please have a look at my previous post Understanding containers (part 05): shared files between the host and containers if you are not familiar with how to start it. As well you can do all the same steps below from any other Python environment.

So, I have my container myjupyter01 running. I am connected to the Jupyter UI as described in the previous blog.

Install hana_ml


The Jupyter image I used from the Docker Hub registry was jupyter/minimal-notebook. It contains already some popular data processing packages, like pandas.

But additionally, I need to install hana_ml, which -- in its current version 1.0.8 -- is available on PyPI repository: https://pypi.org/project/hana-ml/.

The command to run the installation is python -m pip install hana_ml, but because I am running it from Jupyter notebook with Python3 kernel, I need to run it with ! at the beginning:
!python -m pip install hana_ml



Obviously, this installation step has to be done only once. No need to rerun it in the same container e.g. when reloading the newest files.

Use pandas to import files with data


Let's import the same three files (confirmed, deaths, recovered) from https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_time_series as Ferry used in his example.
import hana_ml, pandas



# Links updated on 2020-03-22

df_confd = pandas.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv')

df_death = pandas.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv')

df_recvd = pandas.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_recovered_global.csv')



#Links from before March 22nd

#df_confd = pandas.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_19-covid-Confirmed.csv')

#df_death = pandas.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_19-covid-Deaths.csv')

#df_recvd = pandas.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_19-covid-Recovered.csv')



As you can see from the preview of the Pandas dataframe, it lists only countries or provinces with confirmed cases, and every day the new column is added with the latest data from the previous day. Lines are added when the first case(s) is confirmed in the new region.

Use pandas to re-format the data frame


Before persisting the data in SAP HANA, let's:

  1. Remove all date columns except the last one,

  2. Rename the last column from the actual date (like today's 3/10/20 to Confirmed).


df_confd_latest=df_confd.drop(df_confd.columns[4:len(df_confd.columns)-1], axis='columns')



df_confd_latest.columns = [*df_confd_latest.columns[:-1],'Confirmed']


Use hana_ml to persist data in SAP HANA table


Now let me connect to my instance of SAP HANA Express with the user hanaml that already exists there...
cc=hana_ml.dataframe.ConnectionContext('12.34.567.890', 39015, 'hanaml', 'MyPasswordReusedEverywhere')

...and convert the Pandas dataframe df_confd_latest into a HANA dataframe hdf_confd.
hdf_confd=hana_ml.dataframe.create_dataframe_from_pandas(cc, df_confd_latest, 'df_confd', force=True)

Once the HANA dataframe is created:

  1. A physical column table is created in HANA and data from Pandas dataframe is inserted there,

  2. HANA dataframe hdf_confd in Python does not store any data in your laptop, but only points to a table HANAML.df_confd in SAP HANA server memory, and all Python operations on the HANA dataframe are physically exected in HANA db without moving data between the server and a client,

  3. To display the result of any operations, we need to apply collect() method to convert HANA dataframe to Pandas (and as a result to bring data from HANA db server to the local client).



Use DBeaver to check data in SAP HANA...


You might remember me already using DBeaver -- the free database tool supporting SAP HANA -- in my previous post "GeoArt with SAP HANA and DBeaver".

I am using it now again, and indeed I can find the table df_confd in the schema HANAML with all the data from the source Pandas dataframe.


...and do a spatial preview


As the table contains latitude and longitude columns I can visualize impacted countries/states right from DBeaver with the following SQL using Spatial data preview.
SELECT NEW ST_POINT("Long", "Lat"), "Country/Region", "Province/State", "Confirmed" FROM HANAML."df_confd";

I needed to change the map projection to EPSG:4326 to get these points on the map. And DBeaver shows me the rest of the record data when I click on any point.



[Below is the old screenshot from 2020-03-11, which demonstrates as well the different granularity of e.g. US data used at that time]



DBeaver spatial preview is not a full-blown geospatial visual exploration tool. Yet it is good enough to see impacted countries/regions (depending on the granularity in the source files).

Should you be interested to learn more about hana_ml...


... then I would definitely recommend checking Hands-On Tutorial: Machine Learning push-down to SAP HANA with Python by andreas.forster.

HANA ML is a part of the new "Advanced Analytics with SAP HANA" topic for CodeJam events. Unfortunately because of the coronavirus situation, we had to cancel the first one organized by jakob.flaman in Bern this month. Another one is organized by pekala on May 27th in Katowice: https://www.eventbrite.com/e/sap-codejam-katowice-registration-99016299417. Hopefully, the situation gets normal by that time, and we will not need to cancel this one too.



Stay healthy ❤️
-Vitaliy (aka @Sygyzmundovych)
12 Comments
Labels in this area
Popular Blog Posts