Predicting the future isn't magic, it's Artificial Intelligence.

  If you are new to deep learning or stucked like me in the middle of your journey to Artificial Intelligence.  There is no good source as compared to fastai. I like the practical approach of the fastai and the teaching of the Jeremy's Top Down approach. I don't want to write a blog about what are the concepts in lecture 1 because there are so many people in the fastai forum who are wonderful in explaining the concepts. If you have any doubt ask in the forum, There is no such thing as a good question or bad question everyone in the community is helping each other out. No one is going to judge you. I am writing this because you may also face the same problem like me and it may help you in the journey. For each notebook I am going to share my experience and this is about Notebook Lecture 1(Intro). I strongly suggest you watch Jeremy's lectures before going through the notebook. First oscillation in my mind is which platform I should use for the lecture whether it is google c...

Let me tell you something, I was studying machine learning and deep learning for 7 month. In this journey, Whatever i learned it gives me the feeling of emptyness. During the journey, I came across Fastai which is an high level api built on top of the Pytorch.  Many times , I started and end up of not completing the course due to various reason but this time, I was so commited to complete it. I don't want to write this blog for an expert in Fastai or an deep learning expert. This is for the person who are one steps behind me. I will be sharing my experience may be you may face in the journey. This is an Introduction, So i dont want to dump anything which myself feels annoying for the begginers. Speaking from the experience people in the fastai forum is really amazing whatever the doubt you may come across there is always some one to help you in the journey. I begin my deeplearning journey with various online courses dumping the knowledge without knowing what and how they actually u...

After learning the Sci-kit learn library, I started to think What, Where,  How should I start learning deep learning which library should go on, I find myself Tensorflow is easy to begin with because TensorFlow is developed by google and they providing a stable and extensive documentation and so on... While learning Image classification I came to know the power of GPU, Everything comes in colab which is very much handy. Main thing I learned about deeplearning is You can do anything as long as you can convert into object into Tensors (Array of numbers) Classification of 101 image of food data image data is done with EfficientnetB0 model using transfer learning. There I, learned the power of Transfer Learning. TensorflowHub providing various Pretrained model, we can use based on our need.  And there is something called Tensorboard whwere we can the evaluate the various model, we build with deeplearning. I attached the sample image of Tensorboard.  And there are so many reso...

Project Overview: Twitter Sentiment Analysis helps to understand the user experience of the online retailers based on the tweets, which is important for business development and product enhancement.This a online retailer problem so based on the tweets we are giving the rating of each customer about online shopping experience. Sentiment analysis architecture:  Approach for Twitter sentiment analysis of online retailers: Scraping the tweets: Taking the tweets with respect to online retailers such as ebay, meesho, olx india, flipkart, amazon, jio mart.  After the extraction of the data from twitter, put into the data frame with respect to retailers.  Concatenating the data frame of each retailer into a single data frame.  Converting the data frame to CSV file.  Data preparation and data cleaning: Importing the CSV file using pandas. Analyse & understand the nature of the data.  Tweets are not actual text format, it contains no text characters, writing a fu...

Unsupervised learning: Unsupervised machine learning refers to the category of machine learning techniques where models are trained on datasets without labels. Unsupervised learning generally use to discover the patterns in data and reduce high-dimensional data to fewer dimensions.  Here, I did work on some of the clustering algorithm using scikit-learn namely,  KMeans, DBScan, Hierrarchial clustering. Dimentionality reduction and manifold learning Learning the Algorithm: I Personally feel that data cleaning and Preprocessing are challanging than training the model. Once you finished those 80% of your work is done. Then you can play around with different type of machine learning algorithm. Each algorithms are effective on its own ways. I learned the clustering algorithms through the "iris" dataset in seaborn.  Lets see some of my learning phase of unsupervised learning through Visualization: KMeans: DBScan:  Hierrarchial clustering: Dimentionality reduction: Man...

Fb-prophet We can predict the time analysis using fb prophet.Prophet follows the sklearn model API. We create an instance of the Prophet class and the call its fit and predict methods. Even we can mention holidays in the analysis. In this method monthly-milk data is used.  Source code

Introduction: Decision tree in general represent the hierarchical series of binary decision. The decision tree in the machine learning works in exactly the same way and except that we let the computer figure out the optimal structure and hierarchy of decision, instead coming up the criteria manually. In this model, I took the Australian weather dataset for forecasting.  Data Preprocessing: We'll perform the following steps to prepare the dataset for training: Create a train/test/validation split. Identify input and target columns. Identify numeric and categorical columns. Impute missing values. Scale the numeric value. Encode categorical columns to one-hot vector. Data Visualization: Tree is split on the basis of gini index.   Plot is based on the important feature of Weather prediction. Hyperparamter tuning: What we observe is that training model is 99% accuracy and validation set is just above the average, which means machine is memorizing the data in order to increase the a...