What do you do with unreliable data in network analysis?

1 Identify the sources of unreliability

The first step is to identify the sources and types of unreliability in your data. Unreliable data can arise from various factors, such as measurement errors, missing values, incomplete information, outdated records, or human bias. Depending on the nature and extent of the unreliability, you may need to apply different methods to correct, adjust, or eliminate the problematic data. For example, you can use data cleaning techniques to remove duplicates, fix typos, or fill in missing values. You can also use data validation techniques to check the accuracy, consistency, and integrity of your data against predefined rules or standards.

2 Choose an appropriate network model

The next step is to choose an appropriate network model that can handle the unreliability in your data. A network model is a mathematical representation of a network, consisting of nodes, links, and parameters that describe the characteristics and relationships of the network elements. There are different types of network models, such as deterministic, stochastic, probabilistic, or fuzzy, that can deal with different levels of uncertainty and variability in the data. For example, a deterministic network model assumes that the data is fixed and known, while a stochastic network model incorporates random variables and distributions to account for the uncertainty and variability in the data. You should choose a network model that matches the assumptions and objectives of your problem, as well as the availability and quality of your data.

3 Apply robust or resilient methods

The third step is to apply robust or resilient methods to analyze your network model and obtain your desired outcomes. Robust methods are methods that can produce reliable and consistent results even when the data is unreliable or noisy. Resilient methods are methods that can adapt to changes or disruptions in the data or the network. For example, you can use robust optimization techniques to find solutions that are optimal or near-optimal under a range of possible scenarios or uncertainties. You can also use resilient network design techniques to enhance the performance or reliability of your network under various conditions or events.

4 Evaluate the sensitivity and validity of your results

The final step is to evaluate the sensitivity and validity of your results. Sensitivity analysis is a technique that measures how your results change when you vary the input data or the parameters of your network model. Validity analysis is a technique that assesses how well your results match the reality or the expectations of your problem. For example, you can use sensitivity analysis to identify the critical or influential factors that affect your results, or to test the robustness or resilience of your solutions. You can also use validity analysis to compare your results with other sources of information, such as historical data, expert opinions, or benchmarks, or to verify the assumptions and limitations of your network model.

5 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?