Automate your Business with AI

Key Steps of a Machine Learning Project

Understand what you need for a Machine Learning Project.

Machine learning is a branch of data science. There are 7 main key steps :

• Data Mining
• Data Cleansing
• Data Augmentation
• Choosing Machine Learning Algorithms
• Custom Models Training (Optional)
• Evaluation
• Reinforcement Learning (Optional)

We align technology with your business goals. We help our clients analyzing their business and existing data, in order to decide what we can do in the next step. We encourage clients to get insights from their offline data with the help of machine learning.

Data Mining

Discovering and crawling useful data

Data Cleansing

Removing incomplete, incorrect, inaccurate dirty data

Not all data crawled from the source is high-quality data for analyzation. Dirty data has a bad inference to the prediction results. By ensuring data quality, we prepare valid dataset to prevent potential errors in the later stages.

Data Augmentation

Creating variations of data from existing valid dataset

An easy to understand example: cropping, translating, flipping, tuning the brightness and contrast of an given image. This helps increasing the sample size and let the computer understands the context of the image despite the noise from the environment.

Machine Learning Algorithms

Picking the suitable algorithm

By understanding your business goal, we help to propose suitable approach to tackle the problem. We may propose more than one approach to the problem to increase the chance getting meaningful insights.

Custom Models Training

Need specific model for custom needs? We can help you to train your own model with provided dataset

If existing models is not a good fit for your problem, we are able to train a custom model for you. We worked with a zipper puller manufacturer to build a puller classification model with provided puller images. This option provides flexibility for solving custom business problems and make the best use of your data.

ML model Evaluation

Test against the prediction

Test against the prediction After integrating existing or building customer machine learning models, we will make sure the accuracy is up to standard for use.

Reinforcement Learning

Active learning from production data, improving the existing Machine Learning ability

Coming soon