Get Started and Make Progress in Machine Learning
转自：http://machinelearningmastery.com/get-started-make-progress-machine-learning/#comment-270064
Why Get Started in Machine Learning
I believe machine learning is an important and fascinating field.
- You can make a difference in problem domains that affect a lot of people.
- Machine learning adds more than a new tool to your toolbox, it gives you a superpower.
- There is nothing like it, algorithms that learn from data.
- The time is ripe given computing power, data availability, demand for skills and the growing community
What is Machine Learning
I believe machine learning is a tool required for solving a specific type of problem.
- Some problems are not tractable with hand-crafted programs.
- Machine learning methods are those that automatically create programs from data in service problems.
- Applied machine learning requires an adjustment in thinking from designing a solution to evaluating candidate solutions.
- It is a field comprised of data, tools and algorithms and only good process can take you from problem to solution.
How to Get Started and Make Progress
I believe learn machine learning is learned best by doing.
- Use standard datasets and an off-the-shelf tools in projects and deliver a useful predictive model for the data.
- Keep each project small, 5-10 man hours so that you can complete it within a week of nights.
- Study different datasets, algorithms, and tools and study other peoples completed projects.
- Build a portfolio of completed projects that you can build upon and demonstrate your growing capabilities.
- Follow a step-by-step process for working a problem end-to-end and refine that process on each iteration.
But, What If…
I believe you can learn and apply machine learning, right now.
- Stop preparing to study machine learning and start a machine learning project.
- Machine learning is just another technology that you can pick-up and apply to solve complex problems.
- Skills in programming and mathematics can make some tasks easier, but are not required to get started and solve problems.
- Applied machine learning is a meritocracy, where your ability to achieve results matters more than your background.
- It is a job of creativity that is not the exclusive domain of applied mathematicians.
