https://markokello.com/blog en-us () () Mon, 20 Dec 2021 00:00:00 GMT https://markokello.com/blog/experimentation https://markokello.com/blog/experimentation Mon, 20 Dec 2021 00:00:00 GMT () StatisticsProduct Analytics https://markokello.com/blog/exceptional-data-scientist https://markokello.com/blog/exceptional-data-scientist This post shares insights and advice on how to become an exceptional engineer. I offers new propositions from the previous interview with Zindi Africa. Mon, 06 Jul 2020 00:00:00 GMT () Learning https://markokello.com/blog/probability-distributions https://markokello.com/blog/probability-distributions Fri, 15 May 2020 00:00:00 GMT () MLStatisticsMathsDeep Learning https://markokello.com/blog/sql-series-1 https://markokello.com/blog/sql-series-1 SQL is very easy but its also very easy to shoot yourself in the foot with grouping/aggregations, the HAVING clause, handling NULLs, and subqueries. I aim to highlight some of the common mistakes and how you can avoid them in this article. Mon, 17 Feb 2020 00:00:00 GMT () SQL https://markokello.com/blog/model-evaluation-online-metrics https://markokello.com/blog/model-evaluation-online-metrics Here we explore online metrics for evaluating models in production. From acquisition to revenue, we look into the AARRR framework to understand the key metrics that drive product success. Sun, 26 Jan 2020 00:00:00 GMT () Product Analytics https://markokello.com/blog/probability-theory https://markokello.com/blog/probability-theory TDA. Tue, 17 Dec 2019 00:00:00 GMT () MathsStatisticsML https://markokello.com/blog/gradient-descent https://markokello.com/blog/gradient-descent G.D is an iterative optimization algorithm for training Machine Learning models with the primary purpose of finding the optimal parameters (weights and biases) for the model. The gradient of the loss function, also known as a vector of partial derivatives, indicates the steepest direction of increase. It is then repeatedly updated by taking a step in the opposite direction. The size of the step is controlled by the learning rate. And through this process, the algorithm gradually drives the loss value lower until it converges toward a local minimum. Tue, 19 Nov 2019 00:00:00 GMT () MLDeep Learning https://markokello.com/blog/derivatives https://markokello.com/blog/derivatives . Wed, 30 Oct 2019 00:00:00 GMT () MLMathsDeep Learning https://markokello.com/blog/information-theory https://markokello.com/blog/information-theory Entropy measures the level of uncertainty or randomness in a dataset. Information gain, in turn, evaluates how effectively a decision tree split reduces this entropy. It measures the reduction in uncertainty achieved by a particular split, helping to identify which features create the most meaningful divisions in the data and lead to better classification decisions. Mon, 23 Sep 2019 00:00:00 GMT () MLDeep Learning https://markokello.com/blog/Linear-algebra https://markokello.com/blog/Linear-algebra This blog explains the mathematics and theory behind key classical machine learning algorithms: Linear Regression, Logistic Regression, k-NN, Naive Bayes, and Decision Trees. Wed, 28 Aug 2019 00:00:00 GMT () MathsStatisticsDeep Learning https://markokello.com/blog/numerical-integrals https://markokello.com/blog/numerical-integrals Computers use numerical methods to estimate integrals because real-world data is often discrete, and general-purpose algorithms are required to handle arbitrary functions. This blog covers deterministic methods for 1D integration, such as the trapezoidal and Simpsons rules, as well as Monte Carlo methods for high-dimensional or complex domains. Sat, 17 Aug 2019 00:00:00 GMT () Maths https://markokello.com/blog/classical-ml https://markokello.com/blog/classical-ml This blog is a follow-up to a presentation I gave at Outbox Hub, Its self-contained and explains the mathematics and theory behind key classical machine learning algorithms. Tue, 06 Aug 2019 00:00:00 GMT () MLStatistics