87 Requirements to Include

EBOOK

in Your Predictive Analytics

and Machine Learning RFP

©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku 111

©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku 1
 Introduction
Introduction
Whether you’re a small team of data scientists or a large organization trying to make
Whether you’re aaccessible
data analysis small team of to
even data scientists
those or a largeroles,
in non-technical organization trying
the tool you to make
choose
data analysis
will accessible
help define how youeven
workto with
those in non-technical
data now and in theroles,
future.the
Thetool you choose
balance you
willstrike
help between
define how you work with data now and in the future. The balance
addressing your immediate needs versus anticipating your future you
strike between addressing
requirements your
will play a big immediate
role in creatingneeds versus anticipating
demonstrable, sustainableyour future
value.
requirements will play a big role in creating demonstrable, sustainable value.
Immediate needs depend upon your team, your technology, your industry, and
Immediate needs depend upon your team, your technology, your industry, and
such as the coding languages they use and the techniques they are familiar with.
Technologies
such range
as the coding from data
languages lakes
they usetoand
visualization solutions,
the techniques theyand
arenew technologies
familiar with.
Technologies range from data lakes to visualization solutions, and new technologies
agencies weigh the role of policy in various sectors of the economy. And your
specific organization has its own processes and use cases that will drive adoption
agencies weigh the role of policy in various sectors of the economy. And your
of and value creation from data analytics.
specific organization has its own processes and use cases that will drive adoption
of and value creation from data analytics.

new technologies will emerge, and will your existing technology be compatible with
them? What will the regulatory and economic environment look like? And how will
newyour
technologies will emerge,
own organization’s andand
culture willprocesses
your existing technology be compatible with
evolve?
them? What will the regulatory and economic environment look like? And how will
your
Weown organization’s
have culture
created this data and processes
analytics evolve?
RFP template to save you time and also to
suggest some evaluation criteria on versatility and scalability. Feel free to add and
We subtract
have created this datainanalytics
requirements accordanceRFPwith
template to save
your needs andyou time andAnd
preferences. alsodon’t
to
suggest some
hesitate evaluation
to reach out to criteria on ifversatility
us directly you have and scalability.
any questions Feel any
about freeoftothe
add and
lines
included
subtract here.
requirements in accordance with your needs and preferences. And don’t
hesitate to reach out to us directly if you have any questions about any of the lines
included here.
1
2 ©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku
Vendor
VendorProfile
Profile
TheThe
vendor profile
vendor provides
profile provides the
theessential
essentialfacts
factsand
and context aboutthe
context about thevendors
vendors
youyou
areare
evaluating. WeWe
evaluating. emphasize
emphasizethe theability
abilityof
ofthe
the vendor providededicated
vendor to provide dedicated
customer
customer support,
support, which
which is isespecially
especiallyimportant
importantin in the
the still-nascent
still-nascent world
worldofof
advanced
advanced data
data analytics.
analytics.

AnalyticsRFP
Analytics RFP
This section contains the technical details of the RFP. Below, we will briefly describe the
This section contains the technical details of the RFP. Below, we will briefly describe the
sections. For individual line items, please refer to the spreadsheet.
sections. For individual line items, please refer to the spreadsheet.

Data Access
Data Access
Database Evaluate the breadth of connectivity of
Database
Connectivity Evaluate the breadth
the solution, includingoftraditional
connectivity
dataof
Connectivity the solution,
sources, newincluding traditional
data sources (NoSQL),data
and
cloud. new data sources (NoSQL), and
sources,
cloud.

Data Transformation Evaluate the ability of the solution to

andTransformation
Data Preparation prepare and
Evaluate the transform data.
ability of the This includes
solution to
and Preparation parsing, normalising, and aggregating
prepare and transform data. This includes
data. Evaluate how easy it is to enrich
parsing, normalising, and aggregating
existing data with first-party or third-party
data. Evaluate how easy it is to enrich
sources.
existing data with first-party or third-party
sources.

Data Discovery Evaluate the ability to search and discover

data with the product
Data Discovery Evaluate the ability to search and discover
data with the product

2
©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku 3
 Coding and Extensibility
Coding and Extensibility
Programming Evaluate the product’s language support
Languages Support
Programming both in terms
Evaluate of breadth
the product’s of languages
language support
Languages Support supported
both in terms of breadth ofsupport
and quality of for each
languages
supported and quality of support for each

Development Evaluate the ability of the platform to

Environment
Development support
Evaluateand
themaintain code.
ability of the This includes
platform to
Environment every possible
support aspect code.
and maintain of modern Integrated
This includes
Development
every possibleEnvironment:
aspect of modernerrorIntegrated
Development
detection, Environment:
completion, error etc.
versioning,
detection, completion, versioning, etc.

Enterprise-Wide
Enterprise-Wide Adoption
Adoption
The goal of this section is to evaluate the overall ability to deploy and
The goal of this section is to evaluate the overall ability to deploy and
achieve adoption of the product among a team.
achieve adoption of the product among a team.

Usability byby
Usability Evaluate
Evaluatethe
the overall abilityof
overall ability ofthe
theproduct
product
Business
BusinessUser
User totobe
beused
used by
by business analysts.The
business analysts. The
evaluation comprises evaluating
evaluation comprises evaluating the the
product
productdocumentation,
documentation, useruserexperience,
experience,
and
andthe
theoverall
overall learning curveto
learning curve tobe
be
expected
expectedwith
with the product.
product.

Project
Project Evaluatethe
Evaluate the ability
ability of
of the
theproduct
producttotocreate
create
Management
Management andmaintain
and maintain projects
projects and
andcollaborate
collaborate
withinaaproject
within project as
as aa team.
team.This
Thisincludes
includes
the ability to document a project,reuse
the ability to document a project, reuse
components, publish results, and organize
components, publish results, and organize
overall collaboration and interactions
overall collaboration and interactions
among team members
among team members

4 ©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku

 Data Exploration and Visualisation
Simple and Evaluate the ability to navigate within a
Advanced Data dataset, as well as to perform statistical
Exploration analysis of the dataset, in terms of data
quality, data distribution, etc.

Charting and Evaluate the ability to create charts and

Reporting reports within a dataset. Evaluate ease
of use of charting, overall quality, and
completeness in terms of number and
complexity of charts.

Interactive Evaluate the ability to create more

and Advanced interactive and advanced visualization,
Visualizations including graph analytics, fractals
charts, etc., that possibly leverage
Javascript frameworks and customization
capabilities.

Machine Learning
Algorithm Evaluate the completeness of the platform
Completeness in terms of algorithms and types of data
supported for machine learning.

Feature Evaluate the capabilities in terms of

Transformation feature transformation suitable for
machine learning algorithms. This includes
possibly numerical transformation, text
transformation, etc.

Model Evaluation Evaluate the ability to evaluate a model

and Interpretation and interpret its performance. This
includes contributing variables analysis,
technical performance analysis, and
simulation

Model Optimization Evaluate the ability of the solution to

automatically optimize a model by
means of selecting the best feature
transformation, selecting the best
parameters for machine learning
algorithms, and selecting + combining the
algorithms.

©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku 4 5

 Natural Language Evaluate the capabilities of the solution for
Processing applying machine learning techniques to
text. This includes text classification use
cases as well as possibly more advanced
Natural Language Evaluate the capabilities of the solution for
NLP use cases (such as sentiment analysis).
Processing applying machine learning techniques to
text. This includes text classification use
cases as well as possibly more advanced
Vision Evaluate the capabilities
NLP use cases of the solution
(such as sentiment for
analysis).
Image/Vision use cases.

Vision Evaluate the capabilities of the solution for

Image/Vision use cases.

Operationalization
Operationalization
Data Workflow Evaluate the ability to deploy, schedule,
Automation and monitor data creation workflows.
This includes the ability to integrate with
Data Workflow Evaluate
third partythe ability to
systems anddeploy,
checkschedule,
overall data
Automation and monitor data creation workflows.
consistency.
This includes the ability to integrate with
third party systems and check overall data
consistency.
Model Lifecycle Evaluate the ability of the solution
Management to manage model evolution in terms
Model Lifecycle ofEvaluate
versioning and evolution
the ability of model
of the solution
Management performance.
to manage model evolution in terms
of versioning and evolution of model
performance.
Model Deployment Evaluate the ability of the platform
to deploy a model in various scenarios
Model Deployment (batch or real-time).
Evaluate the ability ofEvaluate the ability
the platform
totoembed
deploy aa model inin various
variousscenarios
applications,
as(batch or real-time).
well as the ability Evaluate thedeploy
to securely ability
to embed a
and monitor it. model in various applications,
as well as the ability to securely deploy
and monitor it.

6 ©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku

 Solution Architecture
Deployment Evaluate the vendor’s ability to support
Options multiple deployment options, including
laptop, on-premise server, and on-premise.

Scalability Evaluate the scalability of the solution

itself and the ability to sustain high growth
in number of users or overall usage.

Cluster (Hadoop, Evaluate the ability of the solution to

Spark, Docker) integrate with the Hadoop, Spark, or Docker
Integration ecosystem and to leverage existing clusters

Extensibility / 3rd Party Integration

3rd party Evaluate the capability of the platform to
integration connect to third-party applications. This
includes third-party applications such
as data sources, data processing, and
processing. This also includes evaluating
the size of their currently supported third-
party ecosystem.

Plugins and Evaluate the ability to easily extend the

Extensibility platform. This includes the ability to add
new connectors, to analyze and audit the
platform, and to pre-package processing

Security & Governance

Authentication, Evaluate the ability of the solution to
access control, provide fine-grain security and group-level
network, governance security. This could include the ability
to mirror and respect the security of the
various data sources and systems accessed
by the platform.

Auditability Evaluate the ability of the platform to

provide trace and audit logs for the actions
performed by users.

©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku

6 7
 Your Path to
Enterprise AI 300+
CUSTOMERS
Dataiku is one of the world’s leading AI and machine
learning platforms, supporting agility in organizations’
data efforts via collaborative, elastic, and responsible AI, 30,000+
ACTIVE USERS
all at enterprise scale. Hundreds of companies use
Dataiku to underpin their essential business operations
and ensure they stay relevant in a changing world. *data scientists, analysts, engineers, & more

1. Clean & Wrangle 5. Monitor & Adjust

Network_dataset Test Test_Scored

Netezza
Teradata Train MLlib_Prediction

Oracle Vertica
HDFS_Avro Joined_Data

Amazon_S3 HDFS_Parquet

Cassandra 4. Deploy
2. Build + Apply to production
Machine Learning

3. Mining
& Visualization

8 ©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku

©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku 9
 GUIDEBOOK
www.dataiku.com
10 ©2020 Dataiku, Inc. | www.dataiku.com | contact@dataiku.com | @dataiku