State of Machine

Learning and Data

Science 2020

Enterprise Executive Summary Report

Table of Contents
Overview 02

Key Results 03

Data Scientist Profile 04

Education 07

Data Science & Machine Learning Experience 09

Employment 11

Technology 18

Conclusion 28

2020 Enterprise Executive Summary Report Table of Contents 1

Overview
For the fourth year, Kaggle surveyed its community of data
enthusiasts to share trends within a quickly growing field.
Based on responses from 20,036 Kaggle members,
we’ve created this report focused on the 13% (2,675
respondents) who are currently employed as data
scientists.

We can see a clear picture of what is common in the

community but also the diverse attributes of its members.

Report
Methodology
The content of this report focuses on respondents who are
currently employed and chose their current job title as
“data scientist”. There are many other job titles that
support data science and machine learning workflows and
you can find their responses in the complete 2020 survey
dataset on Kaggle.

Many survey questions were multiple choice with the

ability for respondents to select all options that applied to
them. For that reason, you may see visualizations where
the total percentage is more than 100%. All monetary
amounts captured in the report are in USD.

2020 Enterprise Executive Summary Report Overview 2

Key Results
Profile
Data science continues to have a heavy gender
imbalance, with most identifying as male

The vast majority of data scientists are under 35 years

old

Over half of data scientists have graduate degrees

Education and Employment

Most data scientists continue to learn outside of
formal education

Most data scientists have been coding for less than a

decade

More than half of data scientists have less than three

years of experience with machine learning

Data scientists in the United States make substantially

more money than their international counterparts

Technology
More data scientists use cloud computing compared to
2019 results

Scikit-learn is the most popular machine learning tool

in 2020, with over four in five data scientists using it

Tableau and PowerBI are the most popular business

intelligence tools

2020 Enterprise Executive Summary Report Key Results 3

Data Scientist Profile
Gender
Data science is still suffering from a large gender gap in the
workplace, as 82% of users identify as men. This is only a
slight change from last year’s results, where 84% of users
identified as males. This is the first year we’ve
differentiated between “Nonbinary” and “Prefer to
self-describe,” with each answer coming in around a third
of a percent.

G e n d e r i d e n t i t y o f d ata s c i e n t i s t s

Man 81.9%

Woman 16.4%

Nonbinary 0.3%

Prefer not
to say 1.1%

Prefer to
self-describe 0.4%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

2020 Enterprise Executive Summary Report Data Scientist Profile 4

Age
Similar to 2019 results, data scientists tend to be in their
late 20s or early 30s, with about 60% between 22 and 34.
Only one in five professional data scientists are 40 or older.
There are signs of the numbers skewing even younger, as
generation Z gets more involved. Nearly 7% of data
scientists are aged 18-21, an increase from last year’s 5%.

Though not included in this chart, responses from students

have also increased each year (26.8% in 2020, 21% in 2019,
22.9% in 2018). As these students graduate into the
workforce, we may see future surveys with even younger
data scientists.

A g e r a n g e s o f d ata s c i e n t i s t s

0-17

18-21 6.9%

22-24 13.7%

25-29 25.2%

30-34 20.1%

35-39 13.4%

40-44 8.7%

45-49 5%

50-54 3.1%

55-59 1.5%

60-69 1.8%

79+ 0.6%

0% 10% 20% 30% 40%

2020 Enterprise Executive Summary Report Data Scientist Profile 5

Country
Two countries have far more representation in the Kaggle
community. India makes up almost 22% of Kaggle data
scientists, while 14.5% reside in the United States. Brazil is
a distant third, at under 5%.

M o s t c o m m o n n at i o n a l i t i e s

21.8%
20%

15% 14.5%

10% 6.7%
4.6%
4.2%
5% 3% 3.3%
2.8% 2.8%
2.1% 2.4% 2.6%
1.4% 1.5% 1.8%
0%

Un
Po

Au

Tu

Ca

Sp

Nig

Ge

Jap

Fra

Ru

Bra

Ot

U.S

Ind
ite
rke

he
ss
la

ain
str

na

rm

ia
eri

zil

.
an

A.
dK
nd

ce

ia

r
da
alia

an
y

ing
y

do
m
Responses per country

# of respondents

300+

150

100

50

2020 Enterprise Executive Summary Report Data Scientist Profile 6

Education
Higher Education
Graduate degrees continue to be the norm for data
scientists, with over 68% having obtained either a Master’s
or doctoral degree. Fewer than 5% of data scientists have
no degree beyond a high school diploma.

E d u c at i o n l e v e l o f K a g g l e d ata s c i e n t i s t s

No formal
education past 0.6%
high school

Some
college/university 2.4%
study without earning
a bachelor’s degree

Bachelor’s
24.2%
degree

Master’s
degree 51.1%

Doctoral
degree 17.2%

Professional degree
3.2%

I prefer
not to 1.3%
answer

0% 10% 20% 30% 40% 50% 60%

2020 Enterprise Executive Summary Report Education 7

Ongoing Learning
Data science and machine learning are quickly changing, Coursera, Udemy, and Kaggle Learn top the most common
so it’s no surprise over 90% of Kaggle data scientists mediums in our survey. Unsurprisingly, many Kaggle data
maintain ongoing education. While about 30% take scientists chose multiple resources in the survey, with an
traditional higher education courses, many more learn average of 2.8 mediums selected.
through online materials.

Popular ongoing learning resources

Coursera 62.9%

Udemy 34.7%

University
Courses(resulting in 30.8%
a university degree

Kaggle Learn
30.1%
Courses

DataCamp 29.6%

edX 22.5%

Udacity 19.2%

LinkedIn Learning 12.3%

Fast.ai 11.8%

Other 9.9%

Cloud-certification
programs (direct 9.1%
from AWS, Azure,
GCP, or similar)

None 7.4%

0% 10% 20% 30% 40% 50% 60% 70%

2020 Enterprise Executive Summary Report Education 8

Data Science & Machine
Learning Experience
Programming Experience
Most Kaggle data scientists have at least a few years of Compared to the global audience, United States data
experience under their belt. Just over 8% of data scientists scientists have significantly greater programming
have been programming since the 20th century! That’s not experience. In the US, 37% have been programming 10 or
to say there aren’t newcomers, however. Over 9% have more years, versus 22% worldwide.

taken up programming in the last year. Just under 2% of

data scientists claim to have never written code at all.

P r o g r a m m i n g b a c k g r o u n d o f d ata s c i e n t i s t s Global USA

20+ years 8.5%

7.6%

13.3%
10-20 years
19.6%

21.9%
5-10 years
29.2%

27.9%
3-5 years
25.3%

17.3%
1-2 years
7%

9.3%
< 1 years
0.8%

I have never 1.8%

written code
0.5%

0% 10% 20% 30%

2020 Enterprise Executive Summary Report Education 9

Machine Learning Experience
Most Kaggle data scientists are newer to machine learning As with programming, US data scientists have more
than programming. Slightly more than 55% of data machine learning experience than the global respondents.
scientists have less than three years experience. Less than
6% of professional data scientists have been using
machine learning for a decade or more.

M a c h i n e l e a r n i n g b a c k g r o u n d o f K a g g l e d ata s c i e n t i s t s Global USA

20 or more 2.1%
years
5.1%

3.9%
10-15 years
8.6%

13%
5-10 years
19.6%

10.9%
4-5 years
15.3%

12.3%
3-4 years
17.2%

15.9%
2-3 years
15.8%

21.4%
1-2 years
12.1%

17.9%
Under 1 year
5.6%

I do not use 2.7%

machine
learning 0.8%
methods
0% 10% 20% 30%
2020 Enterprise Executive Summary Report Education 10
Employment
Pay
Companies in the United States are most likely to pay in There are trends regionally, such as India where nearly

the six figures, based on these survey results. Global 90% make less than $50,000 USD per year.

companies have lower salary ranges that are more evenly

distributed.

G l o b a l s a l a r y d i s t r i b u t i o n f o r d a t a s c i e n t i s t s

> $500,000 0.5%

300,000-500,000 0.7%

250,000-299,999 0.3%

200,000-249,999 1.6%

150,000-199,999 4.7%

125,000-149,999 4.5%

100,000-124,999 6.8%

90,000-99,999 3.5%

80,000-89,999 3.2%

70,000-79,999 4.2%

60,000-69,999 3.7%

50,000-59,999 4.3%

40,000-49,999 5.5%

30,000-39,999 5%

25,000-29,999 3.5%

20,000-24,999 3.6%

15,000-19,999 4%

10,000-14,999 5.7%

7,500-9,999 2.7%

5,000-7,499 3.1%

4,000-4,999 1.8%

3,000-3,999 1.8%

2,000-2,999 2%

1,000-1,999 4.3%

$0-999 18.6%

0% 5% 10% 15%

2020 Enterprise Executive Summary Report Employment 11

S a l a r y d i s t r i b u t i o n f o r U S - b a s e d d ata s c i e n t i s t s

> $500,000 0.8%

300,000-500,000 3.9%
250,000-299,999 1.4%
200,000-249,999 8.9%

150,000-199,999 21.3%

125,000-149,999 18%

100,000-124,999 18.6%

90,000-99,999 6.9%

80,000-89,999 5.3%

70,000-79,999 4.7%

60,000-69,999 0.8%

50,000-59,999 0.6%

40,000-49,999 1.1%

30,000-39,999 0.3%

20,000-24,999 0.3%

15,000-19,999 0.3%

10,000-14,999 0.8%
5,000-7,499 0.3%
4,000-4,999 0.3%
3,000-3,999 0.3%
1,000-1,999 0.3%
$0-999 5%

0% 5% 10% 15% 20%

2020 Enterprise Executive Summary Report Employment 12

S a l a r y d i s t r i b u t i o n f o r I n d i a - b a s e d d ata s c i e n t i s t s

> $500,000 0.6%

300,000-500,000 0.2%
200,000-249,999 0.4%

150,000-199,999 0.6%

125,000-149,999 1%

100,000-124,999 1.2%

90,000-99,999 0.8%

80,000-89,999 1%

70,000-79,999 1.6%

60,000-69,999 0.8%

50,000-59,999 2.6%

40,000-49,999 3.4%

30,000-39,999 4.5%

25,000-29,999 4.7%

20,000-24,999 6.7%

15,000-19,999 7.3%

10,000-14,999 9.7%
7,500-9,999 5.7%
5,000-7,499 4.9%
4,000-4,999 3%
3,000-3,999 1.6%
2,000-2,999 1.6%
1,000-1,999 4%
$0-999 32%

0% 5% 10% 15% 20% 25% 30%

2020 Enterprise Executive Summary Report Employment 13

Looking at the most common salaries by country, we see
that US companies are more likely to pay higher salaries.
Companies in Germany and Japan follow, with significantly
higher salaries than the other included regions.

M e d i a n s a l a r y f o r d ata s c i e n t i s t s b y c o u n t r y

125,000-
USA
149,999

Germany 70,000-79,999

Japan 40,000-49,999

Russia 10,000-14,999

Brazil 10,000-14,999

India 7,500-9,999

1,000- 5,000- 10,000- 15,000- 20,000- 50,000- 100,000- 125,000- 150,000-

$0-999
1,999 7,499 14,999 19,999 24,999 59,999 124,999 149,999 199,999

2020 Enterprise Executive Summary Report Employment 14

Companies Employing Data Science
The most notable change from last year is that more Large enterprises and small startups are the most common
Kaggle data scientists are working at the very smallest choices of data scientists in this survey. Over half of
businesses, at over 37% (up from 30% in 2019). employers have less than 250 employees. Yet, one in five
work at companies with over 10,000 employees.

C o m pa n y s i z e ( # o f e m p l oy e e s )

0-49
37.3%
employees

50-249
13.7%
employees

250-999
10%
employees

1000-9,999
17%
employees

10,000 +
22%
employees

0% 10% 20% 30% 40% 50%

Data Science Teams

With small companies being most common, it reasons that Over half of data scientists work at companies with five or
the same is true for data science teams, most of which fewer people on the data science team. Teams of one or
could be fed with two pizzas. two are most common (23.25%), but large teams of 20+
come next at 22.93%.

D ata s c i e n c e t e a m s ( # o f e m p l oy e e s )

0 9.2%

1-2 23.3%

3-4 18.8%

5-9 15.1%

10-14 7.4%

15-19 3.4%

20+ 22.9%

0% 10% 20% 30% 40% 50%

2020 Enterprise Executive Summary Report Employment 15

Enterprise Machine Learning Adoption
Machine learning has become more rooted in the Those exploring (or using it to generate insights) remain
companies where Kaggle scientists work. Nearly 31% of about the same. Kaggle data scientists who said they’ve
data scientists claim well-established ML methods, up from recently adopted ML decreased, likely due to more
28% in 2019 and 25% in 2018. entrenched usage.

Machine learning adoption in the enterprise over time 2020 2019 2018

7.8%
I do not know
3.8%

4.9%

6.9%
No (we do not use
ML methods) 5.5%

4.2%

We have well 30.8%

established ML
methods (ie., models in 28.9%
production for more
than 2 years) 26%

23.9%
We recently started
using ML methods (ie., 30.7%
models in production
for less than 2 years) 32.9%

We use ML methods 13.1%

for generating insights
(but do not put 14.5%
working models into
production) 13.2%

17.6%
We are exploring ML
methods (and maybe 16.7%
one day put a model
into production)
19%

0% 10% 20% 30% 40% 50%

2020 Enterprise Executive Summary Report Employment 16

Spending
There’s plenty of money being spent on machine learning Data scientists from the US spend more money in the cloud
and cloud computing products, but not by all data than their global counterparts. There are more than two
scientists. There’s quite a range, with over a quarter of data times the responses for the highest spending level in the
scientists claiming to have spent no money at all, while one US compared to other countries.

in 10 has spent over $100,000 USD in the last five years.

U S v s g l o b a l e n t e r p r i s e s p e n d i n g i n t h e pa s t 5 y e a r s ( $ U S D ) GLOBAL USA

11.6%
$100K+
25.6%

15.1%
$10K-

$99,999 20.8%

21.3%
$1K-$9,999
18.9%

16.3%
$100-$999
11.5%

10.2%
$1-$99
4.2%

25.6%
$0
18.9%

0% 10% 20% 30%

2020 Enterprise Executive Summary Report Employment 17

Technology
Interactive Development
Environments

Jupyter-based IDEs continue to be the go-to tool for data This is the first year it has been separated out from Visual
scientists, with around three-quarters of Kaggle data Studio. The two combined for over 43% this year, versus
scientists using it. However, this has decreased from last under 30% in 2019.
year’s 83%. Visual Studio Code is in the second spot with
just over 33%.

Popular IDE usage

JupyterLab 74.1%

Visual Studio Code 33.2%

PyCharm 31.9%

RStudio 31.5%

Spyder 21.8%

Notepad ++ 19.4%

Sublime Text 15.2%

Vim, Emacs, or
11%
similar

Visual Studio 10.1%

MATLAB 5.8%

Other 5.6%

None 0.7%

0% 20% 40% 60%

2020 Enterprise Executive Summary Report Technology 18

Methods & Algorithms
The most commonly used algorithms were linear and
logistic regression, followed closely by decision trees and
random forests. Of more complex methods, gradient
boosting machines and convolutional neural networks were
the most popular approaches.

Methods and algorithms usage

Linear or Logistic
Regression 83.7%

Decision Trees or
78.1%
Random Forests

Gradient Boosting 61.4%

Machines (xgboost,
lightgbm, etc.)

Convolutional Neural 43.2%

Networks

ayesian Approaches 31.4%

Recurrent Neural
30.2%
Networks

Neural Networks 28.2%

(MLPs, etc.)

Transformer Networks 14.8%

(BERT, gpt-3, etc.)

Generative Adversial 7.3%

Networks

Evolutionary 6.5%
Approaches

Other 4.5%

None
1.7%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

2020 Enterprise Executive Summary Report Technology 19

Python-based tools continue to dominate the machine The fifth place tool, PyTorch, climbed above 30%, up from
learning frameworks. Scikit-learn, a swiss army knife about 26% in 2019.

applicable to most projects, is the top with four in five data

scientists using it. TensorFlow and Keras, notably used in The most popular of the tools added to the survey this year
combination for deep learning, were each selected on is R-based Tidymodels, reaching over 7 percent.
about half of the data scientist surveys. Gradient boosting
library xgboost is fourth, with about the same usage as
2019.

Machine learning framework usage

Scikit-learn 82.8%

TensorFlow 50.5%

Keras 50.5%

Xgboost 48.4%

PyTorch 30.9%

LightGBM 26.1%

Caret 14.1%

Catboost 13.7%

Prophet 10%

Fast.ai 7.5%

Tidymodels 7.2%

H20 3 6%

MXNet 2.1%

Other 3.7%

None 3.2%

JAX 0.7%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

2020 Enterprise Executive Summary Report Technology 20

Enterprise Cloud Computing
There are clearly three big players in cloud computing, and

it’s no surprise who: Amazon Web Services, Google Cloud

Platform, and Microsoft Azure. Notably, more data

scientists are using the cloud overall. In 2019, about 25%

had not adopted cloud computing, which decreased to 17%

in this year’s survey.

Enterprise cloud usage

Amazon Web
48.2%
Services (AWS)

Google Cloud
35.3%
Platform (GCP)

Microsoft Azure 29.4%

None 17.1%

IBM Cloud / Red Hat

5.6%

Other
4.1%

Oracle Cloud 3%

VMware Cloud 2.9%

Salesforce Cloud 1.9%

SAP Cloud 1.8%

Alibaba Cloud 0.9%

Tencent Cloud 0.7%

0% 10% 20% 30% 40% 50%

2020 Enterprise Executive Summary Report Technology 21

Those who use cloud services were asked about specific
products. Compute servers are the most common
products, followed by serverless technologies. One in five
did not name a cloud product.

Enterprise cloud product usage

Amazon EC2 40.6%

Google Cloud 21.7%

Compute Engine

AWS Lambda 21.1%

No/None 20.3%

Azure Cloud Services 19.8%

Amazon Elastic
Container Service 14.4%

Microsoft Azure
Container Instances 12.5%

Google Cloud Functions 12.1%

Google Cloud App

Engine 10.6%

Azure Functions 9.3%

Google Cloud Run 6.1%

Other 3.4%

0% 10% 20% 30% 40% 50%

2020 Enterprise Executive Summary Report Technology 22

Enterprise Machine Learning Tools
Those who use AWS, Google Cloud Platform, or Microsoft Of those with ML usage, Amazon SageMaker was the most
Azure were asked about machine learning (ML) tools in popular answer, followed closely by Google Cloud AI and
particular. Over half of these data scientists do not use ML ML.
in the cloud.

Enterprise machine learning product usage

No/None 55.2%

Amazon SageMaker 16.5%

Google Cloud AI
Platform/Google Cloud 14.8%
ML Engine

Azure Machine
12.9%
Learning Studio

Google Cloud Vision AI 8%

Google Cloud Natural

Language 7.8%

Azure Cognitive
Services 6.4%

Amazon Rekognition 4.3%

Google CLoud Video AI 4.3%

Amazon Forecast 3.7%

Other 2.9%

0% 10% 20% 30% 40% 50% 60%

2020 Enterprise Executive Summary Report Technology 23

Enterprise Big Data
Business Intelligence tools help data scientists visualize
their data, but four in 10 do not use one. The majority do
employ BI, with Tableau as the most popular tool. Microsoft
Power BI and Google Data Studio round out the top three.

D ata s c i e n t i s t u s a g e o f b u s i n e s s i n t e l l i g e n c e t o o l s

None 38.8%

Tableau 33.3%

Microsoft Power BI 27%

Google Data Studio 9.1%

Other 6.4%

Qlik 5%

Amazon QuickSight 2.9%

Salesforce 2.8%

Looker 2.5%

Alteryx 2.1%

SAP Analytics Cloud 2%

TIBCO Spotfire 1.4%

Sisense 1.2%

Einstein Analytics 0.9%

Domo 0.7%

0% 10% 20% 30% 40%

2020 Enterprise Executive Summary Report Technology 24

Regarding databases, there isn't a clear favorite among
data scientists. MySQL was mentioned most often (35.6%),
followed by PostgreSQL (28.86%) SQL Server (24.93%).

D ata b a s e u s a g e b y d ata s c i e n t i s t s

MySQL 35.6%

PostgreSQL 28.9%

Microsoft SQL Server 24.9%

MongoDB 18.7%

SQLite 16.5%

None 15.4%

Google Cloud BigQuery 13.5%

Oracle Database 12.9%

Amazon Redshift 9.3%

Microsoft Azure 9.1%

Data Lake Storage

7.9%
Other

Amazon Athena 6.7%

Google Cloud SQL 5.9%

Snowflake 5.6%

Amazon DynamoDB 5.1%

Microsoft Access 4.2%

IBM Db2 3.5%

Google Cloud Firestore 2.8%

0% 10% 20% 30% 40%

2020 Enterprise Executive Summary Report Technology 25

Automated Machine Learning
As with machine learning overall, many data scientists
(33%) do not use auto ML tools. Google Cloud AutoML saw
gains from last year’s survey, nearly 14% versus 6% in 2019.

A u t o m at e d m a c h i n e l e a r n i n g f r a m e w o r k u s a g e

Google Cloud AutoML 13.9%

H20 Driverless AI 9.5%

DataRobot AutoML 8.4%

Databricks AutoML 6.5%

0% 10% 20% 30% 40%

2020 Enterprise Executive Summary Report Technology 26

Machine Learning Experiments
Among data scientists who use tools to manage machine
learning experiments, TensorBoard is a clear favorite (over
21%). The closest competitor is Weights & Biases, with 6%.
However, the vast majority (68%) of data scientists do not
use special tools to keep track of and manage their ML
experiments.

Usage of machine learning experiment tools

No/None
68.1%

TensorBoard
21.6%

Weights & Biases

6%

Other 5.4%

Trains
3.1%

Neptune 2.3%

Domino Model Monitor 1%

Polyaxon
0.9%

Guild.ai
0.8%

Comet.ml
0.7%

Sacred+Omniboard
0.6%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

2020 Enterprise Executive Summary Report Technology 27

Conclusion
This 2020 edition of the State of Machine Learning and
Data Science includes insights gathered from a survey of
20,036 Kaggle members. Their answers covered
demographic, education, employment, and technology
usage.

The charts and results are culled from professional data

scientists (covering 13% of respondents). There’s even
more to uncover in the most comprehensive dataset
available on the state of machine learning and data science
today.

Kaggle has published the complete dataset of responses

for the community to review, and we’ll run a competition
from November 18, 2020 to January 6, 2021 to learn even
more about data science practitioners in 2020.

2020 Enterprise Executive Summary Report Conclusion 28