Temporal oscillations of species concentration in reactions are prevalent due to the nonlinear and often autocatalytic dependence of reaction kinetics on the species concentrations. Self heating in non-isothermal systems too is frequently responsible for inducing oscillations in the species temporal concentrations. Quantitative knowledge of oscillations in reaction systems is therefore imperative for the design, operation, control and optimisation of reactors. It is also crucial for prevention of thermal runaways and migration to bi-stable steady states. Chaos is a unique nonlinear dynamical phenomenon which induces unpredictable oscillations and extreme critical sensitivity to reaction conditions. The trajectories of state variables of chaotic systems diverge exponentially in the state space. Most trajectories of a chaotic system never return to the same point, although they eventually approach it to an arbitrarily small distance. Examples of chaotic reaction systems are the Belousov-Zhabotinsky reaction, combustion of H2, oxidation of CO on platinum, anaerobic fermentation of ethanol, polymerization of olefins in fluidized bed, conversion of heavy gas oil to olefins etc. Control of chaos in chemical reaction systems is crucial considering the necessity to mitigate the long term unpredictability and non-uniform system performance, e.g. irregular product concentration/selectivity at any given time. Chaos in chemical systems may be controlled by stabilising unstable periodic orbits using temporal perturbations to an accessible process parameter. We present an algorithm for control of chaos in reacting systems which implicitly takes into account limitations associated with infrequent sampling of product concentrations. This algorithm is shown to perform better than the Simple Proportional Feedback (SPF) algorithm when species concentration measurements are sparse and noise corrupted. This two dimensional algorithm (2D) SPF is further improved by proposing an inferential variant, which utilizes only a single easily observable secondary state variable for measurement and control. The state variable used for the algorithm is derived from the time series of the rate of heat evolution data from the reactor. The aim of this work is to demonstrate this novel design technique for the control of chaos in a mutating chemical reaction system. In this reaction model, the dynamic equations are unknown and only a single state variable is available for measurement and control.