Advanced Data Analysis and Visualisation Presentation

Introduction

A strong and popular method for processing and visualizing data to obtain insights and make wise business decisions is data analysis using Power BI (Business Intelligence). Microsoft's Power BI is a service for business analytics that provides a variety of tools and services for data analysis and visualization.

Power BI overview: Power BI enables users to connect to different data sources, convert, and clean the data, then build interactive reports and dashboards and share their findings with others. Here is an overview of Power BI's essential elements and the data analysis procedure:

Section 1:

The data you are referring to is the Breast Cancer Wisconsin dataset, which is commonly used for breast cancer diagnosis classification tasks. To obtain the dataset, you can visit the UCI Machine Learning Repository, which hosts a wide range of datasets for machine learning and data analysis. Here's how you can access the dataset:

In the search bar, type "Breast Cancer Wisconsin (Diagnostic) Data Set" and hit Enter.
You will find a dataset entry with a description. Click on the dataset title, and it will lead you to a page with more information about the dataset, including the option to download it.

Regarding your other questions

Rationale for Choosing the Data: The Breast Cancer Wisconsin dataset is a popular choice for classification tasks in machine learning, especially for binary classification problems. It's often used for breast cancer diagnosis, where the goal is to classify tumors as malignant or benign based on various features. The dataset is widely available, well-documented, and a good example for practicing data analysis and classification techniques.

Handling Data Issues: You haven't mentioned specific data issues, but in general, data issues might include missing values, outliers, or inconsistencies. Common methods to handle these issues include imputation for missing values, outlier detection and treatment, and data validation and cleansing Data Transformations: Data transformations depend on the specific analysis you want to perform. Common transformations for this dataset may include:

Encoding categorical variables: If there are categorical variables, they may need to be converted into numerical format. Feature scaling: Standardizing or normalizing numerical features to have a consistent scale. Feature selection: Identifying and selecting the most relevant features for our analysis. Splitting the data: Separating the dataset into training and testing sets for model validation.

Section 2:

Use of power query and DAX expressions

Results

This code shows how to use K-Means clustering to put related data points in groups in the breast cancer dataset. It plots the clusters in a 2D space to see the findings after applying PCA to make the data less dimensional. Depending on the demands of your particular dataset and research, you may change the number of clusters and attributes. To conduct a thorough clustering analysis, further work needs be done on data pretreatment and result interpretation.

Decision Tree Accuracy

The accuracy of the Decision Tree Classifier used to classify the breast cancer dataset was about 94.74%. This indicates that the decision tree model can accurately categorise instances of breast cancer using the features that are given.

Decision Tree Classifier

We trained and evaluated a decision tree classifier. The algorithm classified instances of breast cancer as malignant or benign with an accuracy of around 94.74%.

These findings show how effectively the decision tree model handles this categorization problem. It is significant to highlight that model evaluation and further analysis, including feature selection and hyperparameter tweaking, may be carried out to possibly enhance the model's performance.

Findings

High Accuracy with Decision Tree Classifier: The Decision Tree Classifier successfully distinguished between diagnoses of benign and malignant breast cancer with a high accuracy rate of around 94.74%. This suggests that the model works well for categorising instances according to the given attributes.

Useful Trendline Visualisation: The "Radius Mean" vs. "Perimeter Mean" trendline visualisation offered insightful information on the correlation between these two variables. It facilitates the comprehension of relationships and patterns within the dataset.

Cluster Analysis Identifies Possible Cancer Profiles: Using K-Means clustering, several clusters were found in the dataset, indicating the potential existence of various cancer profiles depending on patient variables. This discovery creates opportunities for additional study of various cancer subtypes.

Limitation

Dataset Size and Quality: Both the size and the quality of the dataset that is utilised may have a big influence on the study. It's critical to take into account the data source, data integrity, and if stronger results might come from a bigger dataset.

Model Generalisation: Even if the Decision Tree Classifier performed well on the dataset, it's important to evaluate how well the model applies to fresh, untested data. Its dependability has to be confirmed by additional testing and validation on external datasets.

Feature Engineering and Selection: More sophisticated feature engineering and selection methods would be advantageous for the analysis. Model performance may be improved by choosing the most informative features and developing additional, pertinent features.

Ethical and Privacy Issues: Strict respect to ethical and privacy laws is necessary while handling medical data, especially sensitive patient information. In order to preserve patient privacy and adhere to regulatory obligations, any real-world implementation of these models must take these issues into consideration.

Limited Model Optimisation: Neither hyperparameter tweaking nor in-depth model optimisation were examined in this investigation. Increasing the model's parameters can result in better performance.

Interpretability: While more sophisticated models may not be as interpretable as decision trees, they are still reasonably interpretable. To comprehend the model's decision-making process, take into consideration applying model interpretability strategies.

Correlation vs. Causation: Although the analysis finds patterns and correlations, it does not prove causation. Medical decision-making requires an understanding of the causal links between characteristics and cancer outcomes.

External Validation: To guarantee the practical usability of the results in a clinical environment, the findings should be confirmed on external datasets or through collaboration with medical practitioners.

Data Imbalances: Examine the dataset for any imbalances, since they might impact model performance and call for different approaches.

Regulatory Compliance: When handling medical data, regulatory compliance is a crucial consideration. Verify compliance with pertinent data protection and healthcare laws.

The study demonstrates potential for the detection of breast cancer; nevertheless, these limitations must be addressed to guarantee the validity and practical use of the results. The application of data analysis and machine learning in healthcare must be advanced, and this requires more study, improved data, and ethical concerns.

Conclusion

In conclusion, the classification of diagnoses as malignant or benign has shown encouraging results from the examination of the breast cancer dataset. The Decision Tree Classifier successfully distinguished between these serious medical disorders with a high accuracy rate of around 94.74%. In addition, the use of trendline visualisation and linear regression helped shed light on the relationships between specific variables, which is helpful for identifying possible patterns in data related to breast cancer. The presence of unique clusters in the dataset was also shown using K-Means clustering, offering insight on the probable existence of various cancer profiles based on patient characteristics. This analysis underlines the potential of machine learning and data analysis approaches in enhancing our comprehension and treatment of this important health issue. It also provides a solid platform for future research in the area of breast cancer diagnostics.

Although the results seem encouraging, it's crucial to understand that this study is only the first stage. To confirm the robustness and generalizability of the findings, more machine learning model optimisation, feature engineering, and refining are required. Due to the sensitive nature of medical data, regulatory compliance, patient privacy, and ethical issues are also essential elements in any real-world deployment of such models. The analysis still emphasises the potential for data-driven discoveries in breast cancer research and emphasises the significance of ongoing research in the hunt for better diagnostic and therapeutic approaches.

References

Python Software Foundation. (2021). Python Programming Language. https://www.python.org/

McKinney, W. (2017). Python for Data Analysis: Data Wrangling with Pandas, NumPy, and IPython (2nd ed.). O'Reilly Media.

VanderPlas, J. (2016). Python Data Science Handbook: Essential Tools for Working with Data. O'Reilly Media.

Scikit-learn. (2021). Simple and Efficient Tools for Predictive Data Analysis. https://scikit-learn.org/stable/

Seaborn. (2021). Statistical Data Visualization. https://seaborn.pydata.org/

Hadley Wickham. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer.

Hadley Wickham & Garrett Grolemund. (2017). R for Data Science: Import, Tidy, Transform, Visualize, and Model Data. O'Reilly Media.

Power BI. (2021). Business Analytics | Microsoft Power BI. https://powerbi.microsoft.com/

Microsoft Power BI Blog. (n.d.). https://powerbi.microsoft.com/en-us/blog/

Power Query Formula Language in Power BI. (2021). Microsoft Docs. https://docs.microsoft.com/en-us/powerquery/

Peltier, J. (2018). Data Analysis and Visualization with Microsoft Excel. Wiley.

Chappell, D., & Elman, C. (2017). M is for Data Monkey: A Guide to the M Language in Excel Power Query. Holy Macro! Books.