Transforming Complex Data into Evidence-Based Insights for Decisive Actions, Sustainable Growth and Excellence
Get in TouchWhen you possess valuable datasets and information, our professional statistical consultancy empowers you to utilize them in productive and effective ways. By extracting meaningful knowledge and evidence from your data, we help improve your business outcomes and enhance your scientific, socio-economic, and policy project work. We cater to your needs by elevating your scientific research, business operations, socio-economic studies, policy development, and other projects through the discovery of actionable insights within your datasets. The knowledge we extract creates intelligence and wisdom in your field of work, paving the way for sustainable growth and helping you thrive in competitive environments. Through rigorous analysis, we discover:
• New patterns and trends,
• Significant associations,
• Causal relationships,
• Underlying regularities.
These insights enable you to perform targeted interventions, make accurate predictions, develop robust explanations, and draw reliable conclusions that drive your objectives forward. Partner with us and get our service to transform your raw data and information into strategic advantage and evidence-based decision-making power.
We specialize in providing exceptional support to undergraduate and graduate students with their research projects, theses, and academic essays. Our expertise enables students to elevate their analytical capabilities and produce superior academic work. Our research support services include:
• Instruction and support in advanced statistical methodologies relevant to specific research questions,
• Custom software coding tutorials tailored to unique analysis requirements,
• Step-by-step demonstrations of analytical techniques,
• Guidance on interpreting and presenting research findings effectively.
Our approach emphasizes both theoretical understanding and practical application, ensuring students develop transferable skills that extend beyond their immediate academic projects. We pride ourselves on helping students develop their independent analytical, logical and critical thinking skills while meeting the rigorous standards of academic research.
Are you looking to enhance your statistical and data analytical skills to better manage your business, accelerate scientific and industrial growth, or develop impactful socio-economic projects? We offer tailored educational experiences designed to help you thrive in your field. Our specialized programs include:
• Custom courses and workshops,
• Interactive discussion sessions,
• Targeted seminars and masterclasses.
Curriculum Highlights:
• Exploratory data analysis techniques,
• Advanced regression modeling,
• Predictive analytics and forecasting,
• Clustering and classification,
• Comprehensive risk analysis,
• Uncertainty reasoning,
• probabilistic modeling.
We employ cutting-edge statistical methods, data science frameworks, and machine learning tools throughout our educational offerings. Each program is customized to align with your specific needs and objectives. Choose from flexible delivery options:• In-person immersive sessions,
• Interactive online learning,
• Hybrid approaches.
This flexibility allows you to select the learning pace and schedule that works best for your team. Contact us today to design a customized educational program that addresses your unique analytical challenges.
Comprehensive analysis of your data to identify patterns, trends, and insights that inform strategy and decision-making.
Transformation of complex statistical findings into clear, compelling visualizations that communicate insights effectively.
Development of robust experimental frameworks that ensure valid and reliable results for your research questions.
Development of methodologically sound surveys that gather reliable data and address your specific research needs.
Creation of sophisticated statistical models that describe relationships, predict outcomes, and test hypotheses within your data.
Implementation of advanced causal discovery and inference techniques to find valid causations for useful interventions.
Click on the picture to enlarge
I have developed an effective and operational method to predict accurately, rare or extreme events happening over time using multiple time series data. When it is needed to predict relatively rare events of elevated intracranial blood pressure (ICP) of patients with brain traumatic injuries within the next hour or so from their clinical condition of the last few hours, I developed a probabilistic model to do it with highest accuracy. The idea is not just to use raw data, but extract features from them, e.g., oscillation patterns, spikes, etc. in the selected time intervals of the past. Severe elevations of ICP, that are rare are difficult to predict with traditional prediction models such as neural networks, etc. So, I invented a novel method which calibrate and weigh predictive probabilities in order to select predictions. The method works in such a way that it gives enhanced predictions for the rare events of severe levels of ICP (the ICP level 2) while losing slightly prediction accuracy for normal levels of ICP (the ICP level 1), for the first to the sixth ten-minute time intervals into the future. The weighing of predictive probabilities is done in order to select predictions through minimizing a prediction error loss function that has different losses for different errors..
The model is probabilistic therefore it can incorporate other information types as past ECG signal variation, respiration pattern, blood pressure variation, etc. of the patient. In fact, in this work no such extra knowledge could improve prediction accuracies. But my proposed method can be useful in other cases. This model can be used to predict stock market, exchange rate, interest rate, etc. prediction from multiple sources of data. And it is applicable for prediction of medical conditions of patients with clinical information over the time. And so on.
I have shown that when predictions are done where predicting some events that are minority are more important than the others that are the majority, probabilistic prediction models such as naive Bayesian networks can be constructed in such a way that we get higher prediction accuracies for important events while losing a little bit of prediction accuracy for unimportant events. My novel model (naiveB*) attained 92% accuracy of prediction for companies that are at risk of bankruptcy (EC=1) and 93% of accuracy of prediction for those are not at risk of bankruptcy at all (EC=3). All other models such as neural networks fail to attain such high prediction accuracies for the important events, but work well for unimportant events, which is not desired for investors and other business professionals. This model can be used to predict rare and important events of interest in many business, scientific, etc. contexts..
Using a general form of probability distance metric, namely, Hellinger distance I obtained a generalized version of the Pearson’s correlation coefficient for two ordinal variables such as educational status and socio-economic status of people or similar cases. It can measure any type of stochastic dependence accurately. For continuous variables like age, weight, etc. optimal categorization is needed, as in case of popular maximal information coefficient (MIC). However the measure I obtained is more accurate. And I have shown that any good measure of dependence is computational, even if it has a closed form formula. High level of accuracy is obtained through, on one hand, using an accurate normalization for probability distance used to build the measure, that is the distance between the existing phenomenon of dependence and its assumed independence. And on the other hand, using a general metric distance that is independent of the nature of the dependence. Some researchers have implemented it in R and made available over the internet: wcor: Wijayatunga coefficient. The measure can be useful in genetic association studies, socio-economic studies where sample survey data consist of a lot of categorical variables. .
I specialize in helping businesses and researchers make sense of their complex data through rigorous statistical modeling and analysis, and clear interpretation. With expertise in regression analysis, causal inference, experimental design, survival analysis, prediction, classification, probabilistic modeling, etc. I provide actionable insights that drive decision-making and solve complex problems.
My approach combines technical expertise with clear communication, ensuring that you not only get the answers you need but also understand how to apply them effectively. The knowledge I extract from your data creates intelligence and wisdom in your field of work, paving the way for sustainable growth and helping you thrive in competitive environments.
Partner with us and get our service to transform your raw data and information into strategic advantage and evidence-based decision-making power.
I obtained Ph.D. Degree in Mathematical and Computing Sciences specializing in Statistics and Data Science from the Tokyo Institute of Technology Tokyo Japan in 2007. Ever since I have been involving in statistical research, teaching, analyses and consultancies for real world problems.
Ready to transform your data into insights? Get in touch to discuss how I can help with your statistical needs.
Email: priyantha.wijayatunga@gmail.com
Phone: +46(0)703254262
Location: Umeå, Sweden