I have a data with 1 target variable and 6 predictor variables. It is a time **independent** data as it is a sensor data without any time stamp. The target variable is linear w.r.t the all the predictor variable when plotted individually but the values seems to be in a pattern. Even using the ts() makes the data not usefull as all the six predictor variables differ at the point where they repeat their pattern i.e. say the first variable repeats itself after 15th row and the second after 20th, third variable after 6th row and so on. Hence there is no frequency as such which can give the pattern exactly. All the six variable are highly correlated. Using PCA gives only one component as useful. But my major concern is since the data is repeating itself, the model may be incorrect.

What can be a good approach to model this problem? (Preferably in R).