in Blog

July 21, 2021

13 Tips to remember while working on ML and BI projects [How to Succeed]


Artur Haponik

CEO & Co-Founder

Reading time:

6 minutes

Project management from A to Z is an exciting adventure that will bring a lot of experience both in the professional and personal sphere. Starting a machine learning projects requires a professional approach to project planning and management, which will definitely help to complete the project favorably.

Do you have a new machine learning project or BI project, but you don’t know where to start? Here are 13 tips that will help you avoid a significant extension of project costs. In addition, these tips can help ensure the success of your future machine learning and BI project.

5 key points to remember about BI and Data Warehouse projects

machine learning projects

1. BI tool & User’s requirements

At the beginning of the project, it is very important to choose the right tool for the business intelligence project’s interface. This tool should include key features such as data visualization, visual analytics, interactive dashboards, and key performance indicators. A well-chosen tool will allow you to:

  • collect all essential data
  • eliminate manual tasks
  • reduce business costs

Moreover, at the initial stage, it is very important to know the end users and their expectations. There are many methods for collecting and analyzing requirements. There are discovery workshops, interviews, process models and much more. Collecting and assessing requirements is crucial for the success of the final product.

2. Enterprise Data Warehouse

EDW increased access to the relevant, contextual inter-organizational information, allowing for more intelligent decision-making. This means faster market entry, a higher return on investment (ROI) and greater growth of the company. But it is important to not deploy such a complex system as Enterprise Data Warehouse at once.

Best of all, release data storefronts step by step and check the data quality and business logic inside. Failure to follow this rule may lead to a malfunction.

3. Scalability and Automation

At the beginning of the business intelligence project, you should also keep in mind scalability and automation. To do this, it is best to use continuous integration tools for automating the BI and ETL pipeline. Without automation tools, your solution does not scale and will not be able to be deployed in different environments and scale your system. A system that has system scalability, adoption scalability, data scalability, and analytical scalability is finally more than just business intelligence services.

4. Data Model & Ad-hoc analysis

The data model should always be focused on ad-hoc and olap analysis. Special analysis provides flexibility in an ever-changing business environment and saves money and time. Therefore, use the star schema approach or snowflake method to show your data in a new layer. And remember that to increase query performance and self-service analysis, the data should be denormalized.

5. Machine Learning as part of a modern data warehouse

The last piece of advice about the BI and data warehouse implementation project is to remember machine learning as part of a modern data warehouse.

Nowadays, machine learning is becoming popular in modern data warehouses, which collect large amounts of data from many sources and devices and store them on a single platform for easy search and analysis. When used successfully, machine learning can help ensure infrastructure scalability, cost savings, and flexibility. Also, do not forget to use DW as a data source for your predictive models to optimize the data preparation process and data quality.

To sum up, strategy, analysis, and planning are required for continuous corporate growth and achieving a competitive advantage. The above steps can be very useful for the organization and for the successful planning and launching of business intelligence projects.

8 key points to remember about Machine Learning projects

1. Сhoose the right ML algorithms

In machine learning projects, first of all, it is important to have a deep understanding of business and be able to combine it with knowledge of statistics. In the broad field of machine learning, you need to know exactly what algorithms or methods can be applied to a particular problem and what data will be required for this.machine learning projects
Choosing the right machine learning algorithm depends on several factors: the size of the data, the quality and variety, as well as what answers companies want to get from this data. Moreover, choosing the right algorithm is a combination of business needs, technical characteristics, experiments and available time.

2. Future solutions & Business requirements

When designing machine learning projects, it is important to plan solutions that will also best meet your business requirements, and not just data. Machine learning supports business processes, which means that each solution must exactly meet the business requirements.

3. Data transformation

Transform the data accordingly. The data is transformed to improve its structure and make it better organized. In addition, data transformation facilitates compatibility between applications, systems, and data types.

4. DWH development

When developing DWH dimension tables and facts, keep in mind that some of them will also be used for machine learning processes.

5. Attribute election process

During your machine learning project, you also need to go deeply into the attribute selection process to choose the correct and most essential functions in the right format and structure in the ML process. Otherwise, machine learning models may be built on irrelevant parameters, resulting in inaccurate forecasts.

6. Modeling process plan

You should also correctly design the modeling process so that the model adapts to changing data structures and is scalable per client. A flexible data model will help your company respond and quickly adapt to changes in the internal or external market. Moreover, a flexible data model can connect, rebuild, or redraw lines and workflows to adapt to the new structure after a transaction.

7. Overfitting in the machine learning models

Machine learning models should not be overfit. The overfitted model in real conditions will give worse and unstable results, which can have a big impact on the company.

There are few of the most popular solutions for overfitting the machine learning models:

1. Cross-validation
2. Train with more data
3. Early stopping
4. Regularization

8. Integration plan

The last tip for a successful machine learning project is to properly plan the integration into the system architecture. Remember that before starting a project, you need to plan how ML will be implemented into the system architecture and at what time the entire flow will be executed. Because the entire data production process depends on it.

All in all, if you spend enough time identifying the right question, pre-processing the data correctly, and considering the impact of using a specific model, this can significantly improve the success of your machine learning project.


  • Launching a project requires a professional approach to project planning and management.
  • Machine learning is a part of a modern data warehouse.
  • Strategy, analysis, and planning in business intelligence projects are required for continuous corporate growth and achieving a competitive advantage.
  • In machine learning projects it is important to know what algorithms or methods can be applied to a particular problem.

Addepto is an advanced AI consulting company. We deal with Machine Learning projects and BI projects every day and know how to help companies to succeed in their projects. If you are looking for a trusted partner, don’t hesitate to get in touch with us.


Machine Learning

Business Intelligence