#Datacleaning can be especially problematic in the case of surveys.

Fieldwork companies will typically do some cleaning but, for IP reasons, are often reluctant to share the details of what they have done and on what grounds.

Cleaning is not as simple as flagging "bad" respondents. Many respondents answer some questions quite diligently but others less so. Their answers may reflect their true feelings, though, and are not necessarily "bad."

So, when conducting modeling, statisticians need to think carefully about the questions they may use for a particular model and which respondents (if any) to exclude from the modeling.

Even when the analytics will consist of simple cross tabs, it may not be safe to assume the data are "clean enough." It might be smart to have your statistician check the data before you begin your analysis.

We can also be proactive and alert our fieldwork company in advance regarding response patterns which suggest satisficing, or even include our own traps in the #questionnaire.

✴️ @AI_Python_EN

At present, many statistical approaches are challenging with huge data files because they are computationally intensive.

One consequence is that simple statistical methods and machine learners are commonly used in predictive analytics. Their predictions/classifications are normally good enough - sometimes very good - for decision-making proposes.

However, they can be difficult to interpret and may shed little light on the data generating process (DGP) - the Why.

Fortunately, most statistical methods have been designed to work with samples, some tiny by the standards of data science. It's possible to develop and deploy one model for prediction/classification and another for explanation.

Their predictions/classifications will not correlate perfectly but this is not an issue unless the correlation is poor - there is uncertainty about anything that is unknown! Correlating their predictions can be viewed as a diagnostic.

This "tandem" approach will not always be feasible or necessary but IMO is underutilized.

✴️ @AI_Python_EN

"Learning Problem-agnostic Speech Representations from Multiple Self-supervised Tasks"

Pascual et al.

Paper: https://lnkd.in/dH2j-n2
Code: https://lnkd.in/dRBYYTH

#DeepLearning #AI #ASR #Machinelearning #DNN

✴️ @AI_Python_EN

"Learning Problem-agnostic Speech Representations from Multiple Self-supervised Tasks"

Pascual et al.

Paper: https://lnkd.in/dH2j-n2
Code: https://lnkd.in/dRBYYTH

#DeepLearning #AI #ASR #Machinelearning #DNN

✴️ @AI_Python_EN

I had the great pleasure of having a conversation with Tobias Macey at Podcast init. We discussed the difference between software engineering and machine learning\data science, the ideas behind our http://DVC.org project.
Also, we touch on the topics of open source software and building successful open source businesses. Challenges in growing communities and product management.The podcast is now available to listen to 🎉
https://lnkd.in/gY8ew5Q

#opensourcesoftware #productmanagement #dvc #machinelearning #artificialintelligence

✴️ @AI_Python_EN

Just found very complete github repository about data science



by: Lazyprogrammer

How come lazy programmer build that complete repostory?so I'm lazyer than lay programmer
Github: https://lnkd.in/fcskFXm

✴️ @AI_Python_EN

Hyperbolic Image Embeddings

Khrulkov et al.: https://lnkd.in/e4BQMmM
#PyTorch code:
https://lnkd.in/eS5tisi

#ComputerVision #DeepLearning #PatternRecognition #MachineLearning

✴️ @AI_Python_EN

You can solve a problem with Excel.
You can solve a problem with Python.
You can solve a problem with R.
You can solve a problem with SAS.
You can solve a problem with Hadoop.
You can solve a problem with Tableau.
You can solve a problem with PowerPoint.
You can solve a problem with Word.

These are all just tools.

Don’t get distracted about what everybody thinks is the best tool.

Start with the problem before you start with the tool.

Use what you have, use what you know, and learn as you go.

Because there are many tools out there that can be used to solve the same problem.

#datascience #problemsolving
#tools

✴️ @AI_Python_EN

A must-read for every non-programmer who aspires to become a data scientist ! Check out this article which lists down various tools that anyone with minimal knowledge of algorithms can use to build high quality machine learning models! #RapidMiner #DataRobot #BigML #GoogleCloudAutoML

https://lnkd.in/fgFAq8V

✴️ @AI_Python_EN

Minimal implementation of Deep Dream in TensorFlow 2.0

By Josh Gordon: https://lnkd.in/eacj7KD

#art #aiart #deeplearning #tensorflow #technology

✴️ @AI_Python_EN

I try to organize Cheatsheet that I share, here's the list


1. Data Science Implementation Cheatsheet
https://lnkd.in/fMHtxYP

2. Discovery Analytics Cheatsheet
https://lnkd.in/f396Dqg


3. SQL Cheatsheet
https://lnkd.in/fKyki2j

4. Machine Learning Cheatsheet
https://lnkd.in/fezaQme

5. Docker Chetasheet
https://lnkd.in/ffMrZXj

6. Tutorial Biglist
https://lnkd.in/fyFxQsM

7. Git Cheatsheet
https://lnkd.in/fWSHH_x

8. Self Driving Car
https://lnkd.in/fxMNBEh

9. Heathcare
https://lnkd.in/fn-yWSD

10. Data Science Cheatsheet
https://lnkd.in/fJgruHJ

#machinelearning #datascience #technology

✴️ @AI_Python_EN

Getting started with #NLP using the #PyTorch framework; Building a #RecommenderSystem; Advice for New Data Scientists; All you need to know about text preprocessing for NLP and #MachineLearning; Advanced Keras - Constructing Complex Custom Losses and Metrics; Top 8 Data Science Use Cases in Gaming
https://bit.ly/2X1OW7E

✴️ @AI_Python_EN

Reinforcement Learning Applications in Business


Detail by Yuxi Li
https://lnkd.in/fR7uDNN
#reinforcementlearning #technology #ai

✴️ @AI_Python_EN

Which is the best tool amongst #Python, #R and #SAS for the job? If you are also looking for an answer, then this Infographic is what you should follow. https://lnkd.in/frqar5E

✴️ @AI_Python_EN

Forest Fire Prediction with Artificial Neural Network (Part 2)
https://bit.ly/2G3iOte
#ANN

✴️ @AI_Python_EN

✳️ " All models are wrong, some are useful "

▶️ “A model is a simplification or approximation of reality and hence will not reflect all of reality … While a model can never be “truth,” a model might be ranked from very useful, to useful, to somewhat useful to, finally, essentially useless.”

- Ken Burnham and David Anderson

▶️ “A model which took account of all the variation of reality would be of no more use than a map at the scale of one to one.”

- Joan Robinson

▶️ “The world doesn’t have the luxury of waiting for complete answers before it takes action.”

— Daniel Gilbert

▶️ “Scientists generally agree that no theory is 100 percent correct. Thus, the real test of knowledge is not truth, but utility. Science gives us power. The more useful that power, the better the science.”

— Yuval Noah Harari

♒️ How Do We Know If A Model Is Useful ?


#LearningthroughPN excerpts of things created by others

✴️ @AI_Python_EN

Important Machine Learning algorithms and their Hyperparameters

#machinelearning #datascience #statistics #algorithms

✴️ @AI_Python_EN

THE IMPACT OF OVERFITTING AND HOW AVOID THEM?


12-years-old Tanmay Bakhsi (today 14) tell about Incredibly well spoken manner about what is overfitting and how to avoid them. He told difference about deep vs shallow neural network and difference about how to train them

Qlue: The solution is early stopping algorithm

Interested to know more, Full video with best resolution is avaliable in his youtube channel: https://lnkd.in/fzrFCvU

#deepleaning #neuralnetwork #artificailintelligence

✴️ @AI_Python_EN

Great MLT workshop on TensorFlow.js with Rei and Kai at Google Japan! In case you couldn't get a spot or want to revisit the resources and build a new application you can check out the GitHub repo https://lnkd.in/fnbyT8k

And some more TensorFlow.js examples https://lnkd.in/fVhhzAt

Join us for more Deep Learning workshops! https://lnkd.in/fmhgHMG

#deeplearning #ai #machinelearning

✴️ @AI_Python_EN

Why statistics should make you suspicious
Spiegelhalter on algorithm, luck, bias, probabilities, machine learning and AI.

https://lnkd.in/e-X9hXJ

#artificialintelligence #bias #ai #statistics #ai #bigdata

✴️ @AI_Python_EN

victorzhou.com/blog/intro-to-neural-networks/ An Introduction to

#NeuralNetworks #deeplearning #machinelearning #python

✴️ @AI_Python_EN