1. **Investment decisions** on whole assets, workovers, or infrastructure. Decisions may often be binary with planning aspects of time to invest 2. **Planning events **- Schedule a set of tasks performed by planetary observers These tasks require battery power, and scheduling too many high-power tasks too closely together can result in battery exhaustion or damage to the instruments, even when continuously charging during daylight hours. Carefully scheduled tasks are desired to fulfill mission goals and maximize battery levels. 3. **Well placement** - where to place wells in a discretized field in a sequence demanded by the availability of drilling equipment. The choice of sequence may consider the expected cumulative value at each location, the costs of moving equipment to each new location and the impact of placing too many wells close together 4. **Unsupervised ML x Clustering**: Another example is in the field of unsupervised machine learning. Clustering is a common exploratory data analysis technique that seeks to partition data points into subgroups based on their similarity, generally determined by the [Euclidean distance](https://en.wikipedia.org/wiki/Euclidean_distance) between points in m dimensional space. 1. Data points in the same subgroup or cluster are nearby and therefore more alike than points that are further away and possibly members of other clusters. Such a clustering may define a classification of types and aid in the development of targeted models to predict future outcomes within each subgroup. The partitioning of _n_ data points into _k_ clusters is NP-hard to compute, as the number of partitioning combinations is _k^n_ and therefore grows exponentially with _n_. 2. Algorithms such as k-means have been developed that apply computational heuristics to seek the partitioning with minimum cumulative intra-cluster distances between data points. Often these heuristic based algorithms find local minimum solutions that may miss subtle and important similarities that could improve classification and accuracy of subsequent model predictions.