ML? API? AutoML? Python is the language of choice to solve problems with machine learning, but what can you build in only a few hours? In only a few days? Without any expertise?

A talk to break common misconception with data science projects. Understand data projects with much simpler approach and gain huge gains from it.

Every one says data is the new oil. But do we actually know how to efficiently use it to make our customer lives better, or it’s just another silo of information.

In this talk, we will see a beautiful approach to planning data-based projects inspired by professionals from Google, Twitter, Microsoft, and more. This talk will cover the following things -
1. Planning a data project sprint
2. Establishing purpose and vision.
3. What data matters and what’s trash?
4. Mining the sentiments of users.
5. Diminishing the silos.
6. Tools

From data wrangling to Machine Learning, from Probabilistic Modelling to Optimization, Python provides all the necessary tools to develop a state-of-the-art decision science platform.

During my talk,
1) I will introduce what Decision Science is and why it matters
2) exemplify a Decision Science modelling problem with an example
3) illustrate how Python and open-source Python frameworks can help tackle all the tasks necessary to make an optimal decision

Pandas is an extremely popular Python package for data analysis, including cleaning and visualizing data before passing it to machine-learning models. In many ways, you can think of Pandas as "Excel inside of Python," with a wealth of capabilities that make it easy to work with data. One of the lesser-known parts of Pandas is its handling of dates and times — a crucial part of what many people use in their work. In this talk, I'll introduce the basic capabilities of Pandas when working with dates and times, including: Reading timestamps from external files, calculating time deltas, querying and comparing time data, using time columns as indexes in data frames, and calculating aggregation functions on "time series" data.

Audio Events, or Acoustic Events, are individual distinct sounds. Audio Event Detection (AED) is the task of detecting such sounds, returning precise times that each kind (class) of sound occurs. This can be anything from detecting coffee-beans cracking while roasting, to gunshots on a shooting range, to noise made by construction works - all these are real applications the presenter has developed. This practical talk will show how you can build such a system in Python, using machine learning models applied to audio. The general approaches shown can also be applied to other sensor data such as vibrations, pressure etc.

The use of AI is all-pervasive across industries now. The adoption has picked up pace mostly in the last 5 years and shows no signs of slowing down. This would take us to a world where AI will be involved in making decisions related to almost everything in our lives - from whether an applicant is eligible for a mortgage, if a patient’s scan shows cancer to which route you should take for your commute and which packet of peanut butter you buy based on the search results.

In this talk we make a case that an independent regulator is needed to create the standards and the guidelines for the adoption of the technology across industries. Expecting regulators for specific industries will lead to inconsistent standards and may also leave most of the industries without properly defined standards at best or at worst with no regulatory oversight on how the technology is being used.

ML algorithm are ubiquitous in decision-making process. This talks aims to provide information about uncertainty. We will see how the estimation of confidence in the inference process helps us to take better decisions. This can indirectly enhance the performance of an ML algorithm and helps to have a deeper understanding of the underlying process used in modeling.

Exploratory Data Analysis (EDA) using Python- EDA is way of visualizing, summarizing and interpreting information in the form of Histograms, Box plot, Scatter plot etc that is hidden in raw data. Once EDA is complete and insights are drawn , it's feature can be used for supervised and unsupervised Machine Learning modelling.

Have you ever wondered what happens when a Machine Learning model dies? Where does it go? Is it replaceable? Can we expose their biases and faults (without twitter)? In this talk, I will talk about Machine Learning models lifecycle, how to ensure their health, diagnose when it is time to build and replace an old one, and how to most importantly, resist the urge to 'just tweak it' and deal with training a younger (and hopefully more capable) substitute. Finally, I will discuss best practices for how to replace those systems without shedding a tear, (easy if you are soulless robot), or how to do so in a respectful manner befitting a benevolent creator. knowledge about how to build and deploy Machine Learning models. The aim is to give a general overview without going too deep into technical details, what makes this talk suitable for both beginners and intermediate Python-users.