What a great crowd today at MinneAnalytics DataTech! I hope you all enjoyed my take on AutoML, and please reach out with any questions/feedback. Here is the slideshare deck.
This is a personal blog. Any views or opinions represented in this blog are personal and belong solely to the blog owner and do not represent those of people, institutions or organizations that the owner may or may not be associated with in professional or personal capacity, unless explicitly stated. Read entire disclaimer here.
SMEs (Subject Matter Experts) are critical to the success of Data Science projects, yet you rarely see them getting credit for passing their knowledge along to a Data Scientist. This post is to give them the credit they deserve and give an example of expected time they could put into a successful Data Science project.
Unless the Data Scientist is an expert in the domain (which is rare), they need a SME to work hand-in-hand with, especially at the beginning of a project. They can bring the following assets:
where to get data & tools they use today
nuances in the data
domain (business knowledge), which is how & why things are done a certain way
who are the other key players in the process
An example of a SMEs could be the Marketing Analyst, DBA, Director of Sales, Engineer, and the list goes on. Data Scientists benefit greatly from their help, and whether or not the project is successful, they should be recognized as partner and rewarded for doing so.
If looking to build a ML Model, SMEs are much more involved in initial stages of ML model development, and levels off to 90/10 workload after initial POC. They can help Data Scientists think through scenarios critical for model training and target variables.
However, SMEs are most likely very busy, and carving out time to help Data Scientists are limited, so it’s up the Data Scientist to be prepared, do research, take notes, and clearly communicate. How much time is needed by SMEs? Every project is different, so it really depends. In my experience, if there were such a thing as an ideal project (I wish there was), the time commitment could look something like this:
Again, this is an example of ideal project, but at least a starting point. Thank you SMEs!
This is a personal blog. Any views or opinions represented in this blog are personal and belong solely to the blog owner and do not represent those of people, institutions or organizations that the owner may or may not be associated with in professional or personal capacity, unless explicitly stated. Read entire disclaimer here.
I’ve been working on building AutoML Apps and explaining to others the difference between traditional ML and AutoML. I created visualization below to help tell the story:
Tools and Apps are being built leveraging AutoML, which many are becoming readily available.
This is a personal blog. Any views or opinions represented in this blog are personal and belong solely to the blog owner and do not represent those of people, institutions or organizations that the owner may or may not be associated with in professional or personal capacity, unless explicitly stated. Read entire disclaimer here.
Wanted to thank attendees of last week’s MinneAnalytics conference for attending and providing great feedback to my presentation on leveraging cloud platforms for ETL and creating ML/Data Products. Slideshare deck is available.
This is a personal blog. Any views or opinions represented in this blog are personal and belong solely to the blog owner and do not represent those of people, institutions or organizations that the owner may or may not be associated with in professional or personal capacity, unless explicitly stated. Read entire disclaimer here.
I ended up creating combination Core ML models, in which loosely based on OOD (object orientated design). How this works, is 1st model identifies domain, if greater than 50% probability, it would then call the next model, and if that is greater than 50% probability, which then call the API to get a returned result.
High-level diagram of how iOS can use models to make useful prediction:
Here is the iOS swift function to call both models.
func captureOutput(_ output: AVCaptureOutput, didOutput sampleBuffer: CMSampleBuffer, from connection: AVCaptureConnection) {
// initial model always called for domain detection
guard let model_one = try? VNCoreMLModel(for: imagenet_ut().model) else { return }
let request = VNCoreMLRequest(model: model_one) { (finishedRequest, error) in
guard let results = finishedRequest.results as? [VNClassificationObservation] else { return }
guard let Observation = results.first else { return }
DispatchQueue.main.async(execute: {
confidence_one = Int(Observation.confidence * 100)
})
}
guard let pixelBuffer: CVPixelBuffer = CMSampleBufferGetImageBuffer(sampleBuffer) else { return }
// executes request
try? VNImageRequestHandler(cvPixelBuffer: pixelBuffer, options: [:]).perform([request])
// ADDING 2ND MODEL TO RUN AT SAME TIME AS 1ST MODEL
// model for issue 1
let chosen_all_model = getAllModel(cise: self.Model.cise)
guard let model_two = try? VNCoreMLModel(for: chosen_all_model) else { return }
let request_two = VNCoreMLRequest(model: model_two) { (finishedRequest, error) in
guard let results = finishedRequest.results as? [VNClassificationObservation] else { return }
guard let Observation = results.first else { return }
DispatchQueue.main.async(execute: {
_confidence_two = Int(Observation.confidence * 100)
}})}
In the complete iOS project, I have 12 models available to run based on various context of what the user was doing. The API create on Firebase could handle various combinations of calls based on these contexts, and send back a useful prediction to the User.
This is a personal blog. Any views or opinions represented in this blog are personal and belong solely to the blog owner and do not represent those of people, institutions or organizations that the owner may or may not be associated with in professional or personal capacity, unless explicitly stated. Read entire disclaimer here.