5 Tips to Make Machine Learning Work for Brands
Machine learning is more than the new marketing buzzword. It’s a realm of data management that only leading-edge brands are mastering.
Part of the reason for that is there are still some companies using data analytics tools and labeling them as artificial intelligence, says Tetiana Boichenko yesterday in her article on ItProPortal. She adds that doing machine learning takes time, money and highly skilled personnel. Not all brands can afford that.
In our recent report on marketing technology investment, we found that brands with revenues of $5 million or more were the ones grabbing up the bleeding-edge marketing tech that harnessed data and finessed it into usable insights about customers.
Here are five tips from Boichenko about what marketers can do to get machine learning to work for their brands:
Does Your Brand Really Need Machine Learning?
Marketers can determine first if their organizations need to make predictions about customers responding to marketing, if they need to be able to whittle huge volumes of data into smaller bites that still contain the essential information, or if they need to enhance data security or comply with new laws, like General Data Protection Regulation, Boichenko says.
If Your Brand Does Need Machine Learning, Which of 4 Types Does It Need?
Boichenko runs through the list:
- Supervised is the most common type, possibly 90% of projects use this X+Y method. X is all of the data marketers have on a customer and Y is whether the person is likely to click on a specific ad.
- Unsupervised uses only X, or input data. “Companies use it to find dependencies and patterns people can’t see or notice,” such as communities with similar interests, she says.
- Reinforcement doesn’t involve many business-use cases, Boichenko says. Self-driving cars can use this, she says. “In reinforcement learning, you specify the rules, the environment and the final reward. To get the reward, algorithms try different strategies and learn from their own history and the environment. It is used to select a successive course of steps to maximize the final reward.”
- Deep learning can encompass the other types of machine learning. It’s the closest to human discernment, Boichenko says. It can include natural language processing, speech recognition and computer vision.
Machine Learning Needs Data Preprocessing and Cleaning
Boichenko says data preprocessing involves humans and can’t be automatic.
“You need to merge many data tables and use different data sources. Then data scientists will be able to play with the models and pick which one best fits a specific business objective.”
Is the data good enough for machine learning? Are you including missing data? This is data cleaning, she says.
Choose Between Machine Learning APIs and Your Own Development
Considering 80% of machine learning effectiveness is about properly preparing the data to put into the models, as outlined in the preprocessing step, Boichenko suggests keeping the development in-house allows brands the most flexibility.
She says of machine learning APIs:
“You sacrifice configurability and control of the system. For instance, Amazon service is fixed just on one model – logistic regression. Whereas, ability to choose the right model is one of the prerequisites for the success of a machine learning project. Thus, companies should outsource machine learning to [the] cloud only if they’ve got very specific and simple tasks.”
Hire Machine Learning Experts
Boichenko says the average pay in the U.S. for this role is $127,000 and companies may want to decide which experts they need from this list:
- Data Scientists to deploy the algorithms. They often use off-the-shelf libraries.
- Natural Language Processing Specialists to do something sophisticated with NLP.
- Data Engineers to create Big Data architectures for the project, ensure system scalability and more.
- Computer Vision Engineers if you need image recognition models.
- Machine Learning Engineers if you have to deploy some state-of-the-art algorithms (e.g. for reinforcement learning) and you need to to do some research specific for the project.
- Speech Recognition Engineers if you need some speech recognition done. However, there are not a lot of business cases.
What do you think, marketers?
Please respond in the comments section below.
Related story: 5 Ways AI Is Infusing Fresh Dynamism Into Marketing