Technical support
Quantitative and Qualitative expertise provided to diverse teams ranging from biomedical projects to surveys. The Statistician has experience in local and international collaborations including the Department for International Development (DFID) related projects providing technical expertise to organisations such as IMC Worldwide, UK. The experience also includes project administration for projects spanning across different provinces: recruitment and training of enumerators, transport logistics involving driver recruitment and testing, as well as arranging for accommodation.
Business Analytics: Use data science tools as the most efficient approach to unpack business solutions.
Life Sciences,
Health Sciences, Humanities, Law, Management Studies/
Commerce, Agriculture, Earth & Environmental Sciences
We partner with academic institutions to provide all the statistical data analysis aspects for data driven reports (mini and full dissertations or theses). Whether the datasets are collected through questionnaires, business transactions or specialized instruments in the field, laboratories and hospitals, the fact that they are named "data" implies the need for the science of data to fully understand the truth behind the data. The science of data is a combination of statistics and coding or programming. Without data science tools, the chances are high that the explanation closest to reality about a dataset will likely to remain untold. At Analytics, we have the tools ready for academic researchers.
Engineering &
Science
Engineering, Chemistry and Physics may have software native to these specific fields but where applicable, reshaping a huge amount of data in preparation for that particular software is all about data science.
Mathematics, Statistics and Computer Science are the breeding grounds of data scientists and production lines for the data science tools. During training, hitting a wall is common and we would like to hear from you.
Protocol/Proposal Development Services
Study design
Based on the field of research, nature of subjects, research objectives, study limitations, in line with the ethics committee, etc, advise on the most suitable study design. With the aid of information from previous studies or using scientific assumptions, conduct power analysis for quantitative studies:
- Priori
- Post-hoc
- Criterion
- Compromise
- Sensitivity
A telescopic view of the journey.
Data collection tools
The priority is seamless data collection to enhance quality. Data science tools will handle the data carpentry. We design the data collection tools and advise on the most probable approaches to maximize the quality of data depending on either the subjects under investigation, laboratory conditions, instrumentation or any other research settings:
- Questionnaires
- Interview guides
- Biomedical or laboratory data collection sheets
- Industrial production log sheets
- etc
Setting the GPS coordinates of the destination.
Statistical Data Analysis Plan
Based on the research objectives and the expected data to be collected, advise on the most appropriate statistical methods to be used for:
- Quantitative and
- Qualitative studies
Vehicle selection – journey ideal for aircrafts.
Data Preparation Services
Loading data
Starting point to learning any software. Data comes in different formats. Having challenges or new in loading data into:
- R
- SAS
- STATA
- SPSS
- NVIVO or
- Converting into another format e.g. csv
- etc?
Fuel up.
Data extraction
When formally granted access to the data, at times the variables and records may be in hundreds and from several datasets while only the selection of a few is needed. When sifting a dataset of interest, the use of programming is second to none. May need to:
- Subset variable names or column indices
- Subset value levels or ranges
- Subset data type
- Merge rows or columns
- etc
Assign crew members.
Data exploration
Having selected the variables of interest, the rows may be in thousands, yet an overview of the values in each column is required to:
- Identify erroneous entries
- Assess missingness per column or entire dataframe
- Identify duplicate rows and columns
- Identify columns with constant values
- Detect outliers
- etc.
Pre-flight checks.
Data cleaning
With many variables and too many rows alike, data exploration may suggest the dropping of:
- Unwanted rows or columns
- Specific values in certain columns or data frame
- Characters in specific columns
Clearing runway.
Variable/value transformations
As an extension to data cleaning, creating new variables as may be required to achieve the research objectives or to conform with the statistical method, values may need to be:
- Calculated using a formula
- Re-coded or re-ordered
- Collapsed or concatenated
- Prefixed or suffixed
- Binned: continuous variables
- Dummy variables in Finance
- etc.
Just airborne and taking full control of the data from the cockpit.
Data reshaping
The final product of any data carpentry where a dataset will be ready for dispatch into the next production line to be turned into meaningful information. Any statistical method will require data in at least one of the two:
- Wide format or
- Long format
Final airborne maneuver and setting up the cruising altitude.
Results and Interpretation Services
Descriptive statistics
No hypothesis testing and useful for:
- Smaller sample sizes with low power (by default, suitable for large sample sizes)
- Nested Table1: frequencies, central tendencies, dispersion and positions
- Pareto graphs for column split common to medical data
- Multidimensional plots for visualizing all the variables at once: alluvial, Likert, heatmap, iconplot, correlation plot, parallelplot, slopegraph, etc
- Publication ready – R outputs
Cockpit view
Descriptive + Inferential statistics
Most common as it is now possible to combine data exploration and hypothesis testing – but usually involving at most three variables at a time. In addition to the usual Table1, p-values (and in some cases simple regression equations) are now annotated on:
- Box plots
- Violin plots
- Line graphs
- Scatter plots: linear, loess, gam, etc
- Bar charts: simple, multiple, component, etc
- Pie charts
- Correlation plots
- t-test, Wicoxon test, ANOVA and Kruskal-Wallis
- Fisher’s exact test, Chi-Square test and McNemar’s test
- Cramer’s V and correlation test
- etc
- Publication ready – R outputs
Relying on cockpit view + flight instruments
Inferential statistics
More rigorous and regression-based approaches commonly involving:
- Techniques for variable reduction/selection and detection of influential observations: PLS, VIF, PCA, Reliability, Cook’s distances, etc
- Categorical dependent variable logistic models – binary (non- and time dependent), multinomial (non- and ordinal), etc
- Numeric dependent variable models – Poisson, Gaussian, Gamma families, etc
- Many other model building techniques: mixed models, etc
Other common inferential techniques include:
- Sensitivity and specificity
- Inter-rater
- Case control
- etc
- Publication ready – R outputs
Relying on the flight instruments only.
Interpretation of quantitative results
In the research process, it is a milestone to have managed to extract almost all the patterns in a given quantitative dataset. The results exist as either tables or graphs that are communicating rich technical details. It takes another milestone to compose the story into black and white for the public domain. It is the translation from one language to another.
Passenger and ground crew update.
Qualitative results
Cockpit voice recorder.
Interpretation of qualitative results
With all the participants’ responses re-arranged into convenient paragraphs that clearly group the main themes, writing a coherent story of the emerging ideas brings light to the research questions. It is all about understanding how the algorithm is reshuffling the texts into something meaningful.
Understanding the flight journey.
Training Services
Research project administration
It is easier said than done. The quality of any report starts with collected data. The data in turn rely on the quality of the field workers. However, the field workers also depend on the quality of the data collection instrument. We provide training on:
- Field work – enumerators
- In addition we can also execute the actual project administration. There after take over from the data capturing, analysis and report writing.
Setting the scene.
R programming and more
Currently taking mostly requests for R training. Whether you are a beginner or advanced user, R capabilities are usually the latest and do almost anything you can think of with data in amazing ways.
Beginners will definitely leap forward whilst advanced users may find solutions to their code from the problems we have already solved.
Other statistically related
- EndNote citations
- Data science = Statistics + Coding
Highlighted are the most common requests. We will be more than happy to hear from you.
Business Analytics Services
Data collection tools
As long as some datasets are captured on hard copies or directly into spreadsheets on a routine basis, that is the perfect opportunity to realize the power of data science. We help with the re-designing of both the hard copies and the spreadsheets.
Focus on making the data capturing easy – data carpentry will handle the rest.
Automation of report generation
No matter how tedious it is to generate weekly, monthly or annual reports, as long as the data collection tools are synchronized with our programming code, hours or days’ work is turned into the required report with a single click.
Program the routine and tedious tasks to be automated and focus on what matters.
Other statistically related
Data science = Statistics + Coding
Highlighted are the most common requests. We are more that happy to hear from you.
