Sebastian Vasquez

Hello #datafam !! I want share my last dashboard using Tableau about environmental impacts in México... I love this viz because the toggle buttons give more interaction for the user! Key features: Environmental to level water 🌊 Environmental to level floor 🏠 Environmental to level vegetation 🏕 Tableau Public:https://lnkd.in/djRrC-cY Proverbs 2:6-7 NKJV For the LORD gives wisdom; From His mouth come knowledge and understanding; He stores up sound wisdom for the upright; He is a shield to those who walk uprightly #tableau #tableaudashboard #dashboard #dataanalyst #enviromental #visualization #tableaupublic #mexico #datamexico Tableau

7

🏎️ Desights Formula 1 Data Challenge: Predict Pit Stop Strategies for the Mexican Grand Prix! Ready to take on the world of Formula 1 through data science? This month’s challenge focuses on predicting pit stop strategies for the upcoming 2024 Mexican Grand Prix, powered by Ocean Protocol. You'll get the chance to work with real-world F1 data from past seasons and build machine learning models to predict key race strategy components like: Number of Stints Tire Compounds Laps per Stint Average Lap Time The goal? Optimize race strategies to help teams stay ahead. In a sport where every second counts, your model could be the difference between a podium finish and a missed opportunity. 🔑 Why should you join? $10,000 prize pool for the top 10 participants 🏆 Earn points toward the 2024 Championship for even more rewards 📈 Real-world F1 data to test and improve your skills 🏎️ This challenge is perfect whether you're a seasoned data scientist or just starting out. Ready to take on the high-stakes world of F1 strategy? Submit your model before the deadline and get your engine roaring! 🚀 Join the challenge here! https://lnkd.in/eNYkz-8e #Formula1 #DataChallenge #Desightschallenge

8

Hey #datafam, Have you created any Spain themed Tableau dashboards? Topics can be on anything, as long as the dataset is focused in Spain. We would love to see your creations! Share them with us on the comments below. #spain #dataviz Tableau User Group Barcelona

14

Hi everyone, This post is a follow-up on a weekly series of posts documenting my progress on a personal data science project on demand forecasting. My goal is to develop a demand forecast for hourly energy demand from Spain. I have linked to the original dataset here: https://lnkd.in/ek47Ki8u And I have also linked in to my GitHub with my source code: https://lnkd.in/eeDUBDGF In last week's post, I talked about using an ARIMA model to forecast demand in the first two weeks of February 2015 using the demand from January 2015. I talked about how the ARIMA model was technically better than other forecasts because the errors weren't meaningfully auto-correlated, but the forecast wasn't actually more accurate by usual accuracy measures such as RMSE. Since then, I have tweaked the parameters in my ARIMA model, and that has resulted in a clear improvement. My modified ARIMA forecast now has the lowest RMSE, so it's clearly the best forecast I have experimented with until now. I spent the rest of the week researching different forecast techniques that incorporate different kinds of seasonality. In previous posts, I showed that the demand was seasonal on an hourly, daily, and monthly basis. So far, all the forecasts techniques could only work with one kind of seasonality. Since I've thus far only used January's demand to predict February's, and those two months had similar demand historically, monthly seasonality hasn't been a major issue. But I am trying to find a good way to incorporate daily seasonality since demand tends to be lower on weekends. I used a forecast which incorporated my ARIMA model with a harmonic regression model. The results were very disappointing. This more complicated model performed worse than the simpler ARIMA model, and is one of the worst models I've tested thus far. On the table below, you can see the accuracy measures for the simple ARIMA model (arima200) and the ARIMA + harmonic regression model (arima_harmonic), in addition to other forecast techniques. I think I should test this model over a longer time-frame, but these initial results are not promising. The other issue is that this model is much more complex than anything I have tried until now. Complex forecasts risk not receiving any buy-in from stakeholder. It's also possible I made a mistake in my code when writing this model. That's it for this week. I am glad that I was able to improve my original ARIMA model this much. I hope I can find a better solution for the multiple kinds of seasonality.

6

If you’re in the data science community, you may have come across Andres Vourakis. I don't think he needs much introduction but in case you don't know he is a data scientist who has made incredible contributions to the data science community and has an amazing newsletter, where he inspires data enthusiasts worldwide! I just read his article, "No Money or Job Offers: My Unusual Path to Data Science" and I loved it! His journey from a non-technical job to a successful career in data is a must-read for anyone just breaking in. Here are 3 key takeaways from reading it👇): 1️⃣ Every role is a chance to sharpen your analytical mindset. You don’t need to have the data analyst title to start thinking like and acting like one. Any job—whether it’s customer service, retail, or even administrative work—gives you the chance to develop problem-solving and data-driven thinking. Look at patterns, identify bottlenecks, and think about how data could help solve them. Your first data job is already in front of you; it’s just a matter of mindset! 2️⃣ Build confidence through small wins. Getting your first job in data can feel overwhelming. Start with small, manageable projects—even personal ones. Whether you’re automating a task with Python or creating a simple dashboard in Excel, these small wins build your confidence and can be showcased to potential employers. Small projects can snowball into major opportunities! 💪 3️⃣ Tech skills can be developed in any environment. This takeaway from the article really resonated with me. Andres used a Python script to automate a task at a retail job—saving his company 100 hours a month! No matter where you’re working right now, use your tech skills to solve problems, automate processes, and deliver value. You can make a huge impact right where you are. Remember, you don’t need the perfect role to start thinking, acting, and growing like a data analyst. 🧗 Be Curious and Explore: 🧗 Always seek knowledge and don’t hesitate to ask questions. The our field is vast, and curiosity will drive your learning journey. 💎 Value Over Title: 💎 Focus on the value you can provide rather than your job title. Deliver meaningful results, and people will take notice. 🔍 Seek Opportunities: 🔎 Don’t wait for them to come to you! Be proactive in asking for responsibilities and new challenges. Start where you are, build your skills, and ask for the opportunities today. You can read the same article I read here ⬇️ https://lnkd.in/gKV9VfCA #DataAnalystJourney #CareerGrowth #TechSkills #PythonForData #SmallWinsBigImpact #SelfDevelopment #DataAnalytics #NewToData #LinkedInTips

15

6 comentarios

TradesViz's pivot grid shows pnl, quantity, and count of tags, all explorable on a single page. You can chart and analyze nearly anything: symbols, tags/groups, ranges, vs pnl, r-value, mfe/mae +more! This is the most powerful single-tab analytics dashboard for traders 👇 #stock #trading #futures #options #tradingperformance #tradingjournal #tradesviz #crypto #trader #stockmarket #forex #stocktrading #daytrader #data #datavisualization #analytics

1

🔍 Is Your Data Normally Distributed? Understanding your data distribution is crucial before applying statistical or machine learning models that assume normality. In my latest post, I explore two key methods to check for normality: Shapiro-Wilk Test: Ideal for small datasets. D’Agostino’s K-squared Test: Works well for larger datasets and combines skewness and kurtosis. Both methods give you insight into whether your data fits a normal distribution. I’ve also included Python code examples and a guide to interpreting the results. Learn how to: ✅ Apply normality tests in Python ✅ Interpret p-values to identify outliers ✅ Use visual aids like Q-Q plots to complement your analysis #DataScience #Python #DataAnalytics #MachineLearning #DataVisualization #NormalDistribution #OutlierDetection #DataAnalysis #StatisticalAnalysis #PythonProgramming #DataWrangling #QuantitativeAnalysis #Analytics #BigData

13

3 comentarios

🌟Hello Everyone!🌟 I'm happy to share my new Portfolio Website😀. Contains some EDA 🔍 projects, Kaggle competition 💥 notebooks, and some AI 🤖 projects. Feel free to explore my repositories and Kaggle contributions. 🔗 GitHub: https://lnkd.in/eqKtnncQ 🔗 Kaggle: https://lnkd.in/exZmyU_Q #datascienceportfolio #PortfolioLaunch #DataScience #GitHub #Kaggle #MachineLearning #DeepLearning #ArtificialIntelligence #Math #Portfolio #ARCgames #ClimSim #NaturalLanguageProcessing

10

2 comentarios

Working in tech is challenging and exciting 🚀 . However, at times, it can become overwhelming. So many changes, new technologies, and tight deadlines can take a toll. They drain your energy and make you lose motivation. In those moments when energy is low, I always try to go back to the basics. To revisit the important concepts that made me fall in love with this Tech world and specifically the world of data. DataCamp is a great place to refresh those concepts and get your mind in order. I wanted to share that I have completed the "Introduction to dbt" course. Thanks, DataCamp, for being my trusty guide on this data journey!

20

Sharing my latest project: a population density map of Colombia using WorldPop's 2022, unconstrained data (https://data.worldpop.org/). This is based on the tutorial from Milos Popovic, PhD. I faced some challenges because Colombia's data was in grayscale, unlike the Netherlands' binary (black and white) data. The original code couldn't handle these variations in shades of gray, which represent different population densities. To solve this, I modified the code to interpret these shades correctly. The key was using grid cells to simplify the data complexity. This experience taught me a lot about adapting to different data formats and handling continuous data effectively. Git code: https://lnkd.in/e-wYWgkn Last post: https://lnkd.in/exxJi-Xy #mappingtheworld

42

4 comentarios

Customer Churn Prediction with Machine Learning: Key Insights from Diego Hurtado's Recent Post! A big thank you to Diego Hurtado for his fantastic deep dive into customer churn prediction using machine learning models! 📊 Diego breaks down the importance of predicting customer churn, highlighting that retaining customers is not just about business success—it's about understanding why they leave in the first place. His approach outlines several critical steps in building a successful churn prediction model: Data Collection & Preparation: Collecting comprehensive datasets, such as customer demographics, usage patterns, billing history, and customer service interactions. Cleaning and preparing this data is crucial for effective analysis. Feature Engineering: Identifying the most relevant features—such as customer tenure, service usage, and interaction frequency—can significantly improve model accuracy. Model Selection: From Logistic Regression to LSTMs, Decision Trees to XGBoost, choosing the right algorithm depends on the data's nature and the desired level of accuracy. Diego even demonstrates the power of an LSTM classifier for capturing the sequential dependencies in customer behavior! Model Evaluation & Interpretation: Utilizing Confusion Matrix, ROC Curve, and AUC to evaluate model performance and understand the factors driving churn. Actionable Insights: Feature importance analysis, whether via XGBoost or permutation importance methods, helps businesses target their retention strategies efficiently—focusing on the factors that matter most to their customers. 💡 Why is this Relevant to Data Engineering? 💡 As data engineers, we play a critical role in setting the foundation for such predictive models. We handle the data pipeline's creation, ingestion, transformation, and storage—ensuring that clean, accurate, and timely data is available for machine learning models. Building robust data infrastructure and using tools like Apache Kafka, Spark, and various ETL processes allow data scientists and analysts to focus on modeling and deriving insights. By working together, data engineers and data scientists can create powerful, end-to-end solutions that prevent customer churn and drive sustainable growth. 🚀 Once again, thanks, Diego Hurtado, for your inspiring post! 🙏 For anyone interested in customer analytics or data science, I highly recommend checking out his detailed breakdown and examples. I'll put a link to his article below: https://lnkd.in/etg9ezqE. #MachineLearning #DataEngineering #CustomerChurn #DataScience #AI #DataAnalytics #CustomerRetention #DataPipelines #ChurnPrediction #FeatureEngineering #ModelEvaluation #XGBoost #LSTM #ETL #BigData

1

The demand for Data Scientists is rapidly growing, but what sets the best apart?    We explore the essential technical skills, industry experience, business acumen, and soft skills that make top data scientists so desirable. From mastery of machine learning to the ability to communicate complex insights effectively, discover how the right blend of expertise and responsibility positions data scientists as key players in today’s tech-driven economy. #DataScience #MachineLearning #AI #BusinessIntelligence #DataSkills #DataScientists #TechIndustry #BusinessInnovation #TechTalent #FutureOfWork

6